[time-nuts] Sine to square wave converter Phase Noise
Recently I have been comparing the phase noise of the HOlzworth HX4210 and the LTC7957-4. I have found that the performance of these devices is comparable with a 10MHz sinewave input with a PN noise floor below -160dBc/Hz. For offsets below 100Hz the LTC6957-4 is quieter and for offsets below 50Hz or so is lower than that of my measurement setup. I merely mounted the LTC evaluation board in a diecast metal box to reduce the effect of air currents on the LTC7957-4 and used an ultra low noise 3.3V power supply (Abracon) to minimise the power supply contribution to the measured phase noise. Achieving a phase noise performance equal to that on the datasheet was relatively easy. In fact for low offsets the phase noise is lower than that shown on the datasheet. I've also measured the phase noise of a comparator based sine to CMOS converter and its PN floor is about 20dB higher than that of the LTC6957. On completion of measurements PN plots for the various sine to square conveters tested will be added to the web pages I maintain on Didiers site.. Bruce ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Sine to square wave converter Phase Noise
You alternate LTC7957-4 and LTC6957-4 and it is only the later that exists. The LTC6957-4 has a single-stage amplifier stage input, with somewhat programmable bandwidth. I assume you used the 50 MHz BW setting. Cheers, Magnus On 07/29/2014 10:55 AM, br...@ko4bb.com wrote: Recently I have been comparing the phase noise of the HOlzworth HX4210 and the LTC7957-4. I have found that the performance of these devices is comparable with a 10MHz sinewave input with a PN noise floor below -160dBc/Hz. For offsets below 100Hz the LTC6957-4 is quieter and for offsets below 50Hz or so is lower than that of my measurement setup. I merely mounted the LTC evaluation board in a diecast metal box to reduce the effect of air currents on the LTC7957-4 and used an ultra low noise 3.3V power supply (Abracon) to minimise the power supply contribution to the measured phase noise. Achieving a phase noise performance equal to that on the datasheet was relatively easy. In fact for low offsets the phase noise is lower than that shown on the datasheet. I've also measured the phase noise of a comparator based sine to CMOS converter and its PN floor is about 20dB higher than that of the LTC6957. On completion of measurements PN plots for the various sine to square conveters tested will be added to the web pages I maintain on Didiers site.. Bruce ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Sine to square wave converter Phase Noise
Sorry, I meant the LTC6957-4 of course._SELA = l FILTA = L FILTB = H ie an input stage BW of 160MHz as recommended for a 10MHz + 10dBm input. Input for the HX4210 was +14dBm. Comparator was that used in David Partridge's programmable divider (includes the 74AC04 buffers). Bruce On July 29, 2014 at 5:37 AM Magnus Danielson mag...@rubidium.dyndns.org wrote: You alternate LTC7957-4 and LTC6957-4 and it is only the later that exists. The LTC6957-4 has a single-stage amplifier stage input, with somewhat programmable bandwidth. I assume you used the 50 MHz BW setting. Cheers, Magnus On 07/29/2014 10:55 AM, br...@ko4bb.com wrote: Recently I have been comparing the phase noise of the HOlzworth HX4210 and the LTC7957-4. I have found that the performance of these devices is comparable with a 10MHz sinewave input with a PN noise floor below -160dBc/Hz. For offsets below 100Hz the LTC6957-4 is quieter and for offsets below 50Hz or so is lower than that of my measurement setup. I merely mounted the LTC evaluation board in a diecast metal box to reduce the effect of air currents on the LTC7957-4 and used an ultra low noise 3.3V power supply (Abracon) to minimise the power supply contribution to the measured phase noise. Achieving a phase noise performance equal to that on the datasheet was relatively easy. In fact for low offsets the phase noise is lower than that shown on the datasheet. I've also measured the phase noise of a comparator based sine to CMOS converter and its PN floor is about 20dB higher than that of the LTC6957. On completion of measurements PN plots for the various sine to square conveters tested will be added to the web pages I maintain on Didiers site.. Bruce ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Sine to square wave converter Phase Noise
Bruce, Thank you, that was very useful information. If I ever do a re-spin of the divider I will likely use the LTC6957-3 or -4 instead of the ADCMP600 (I would need interest for at least 40 boards to be able to do this at anything like an economic price). Will you be putting the full details of the circuit you used up on Didier's web site? Did you use a transformer based circuit similar to Design Note 514 or just an input capacitor? Regards, David Partridge -Original Message- From: time-nuts-boun...@febo.com [mailto:time-nuts-boun...@febo.com] On Behalf Of br...@ko4bb.com Sent: 29 July 2014 12:00 To: Discussion of precise time and frequency measurement Subject: Re: [time-nuts] Sine to square wave converter Phase Noise Sorry, I meant the LTC6957-4 of course._SELA = l FILTA = L FILTB = H ie an input stage BW of 160MHz as recommended for a 10MHz + 10dBm input. Input for the HX4210 was +14dBm. Comparator was that used in David Partridge's programmable divider (includes the 74AC04 buffers). Bruce On July 29, 2014 at 5:37 AM Magnus Danielson mag...@rubidium.dyndns.org wrote: You alternate LTC7957-4 and LTC6957-4 and it is only the later that exists. The LTC6957-4 has a single-stage amplifier stage input, with somewhat programmable bandwidth. I assume you used the 50 MHz BW setting. Cheers, Magnus On 07/29/2014 10:55 AM, br...@ko4bb.com wrote: Recently I have been comparing the phase noise of the HOlzworth HX4210 and the LTC7957-4. I have found that the performance of these devices is comparable with a 10MHz sinewave input with a PN noise floor below -160dBc/Hz. For offsets below 100Hz the LTC6957-4 is quieter and for offsets below 50Hz or so is lower than that of my measurement setup. I merely mounted the LTC evaluation board in a diecast metal box to reduce the effect of air currents on the LTC7957-4 and used an ultra low noise 3.3V power supply (Abracon) to minimise the power supply contribution to the measured phase noise. Achieving a phase noise performance equal to that on the datasheet was relatively easy. In fact for low offsets the phase noise is lower than that shown on the datasheet. I've also measured the phase noise of a comparator based sine to CMOS converter and its PN floor is about 20dB higher than that of the LTC6957. On completion of measurements PN plots for the various sine to square conveters tested will be added to the web pages I maintain on Didiers site.. Bruce ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Sine to square wave converter Phase Noise
Further to my previous post: I note that the LTC6957 is 3.3V CMOS rather than 5V. Do you have a recommmendation for a level converter to 5V logic? Would a 74AC04 using a 5V supply cope well enough in that role or is there a better solution? Regards, David Partridge -Original Message- From: time-nuts-boun...@febo.com [mailto:time-nuts-boun...@febo.com] On Behalf Of br...@ko4bb.com Sent: 29 July 2014 12:00 To: Discussion of precise time and frequency measurement Subject: Re: [time-nuts] Sine to square wave converter Phase Noise Sorry, I meant the LTC6957-4 of course._SELA = l FILTA = L FILTB = H ie an input stage BW of 160MHz as recommended for a 10MHz + 10dBm input. Input for the HX4210 was +14dBm. Comparator was that used in David Partridge's programmable divider (includes the 74AC04 buffers). Bruce On July 29, 2014 at 5:37 AM Magnus Danielson mag...@rubidium.dyndns.org wrote: You alternate LTC7957-4 and LTC6957-4 and it is only the later that exists. The LTC6957-4 has a single-stage amplifier stage input, with somewhat programmable bandwidth. I assume you used the 50 MHz BW setting. Cheers, Magnus On 07/29/2014 10:55 AM, br...@ko4bb.com wrote: Recently I have been comparing the phase noise of the HOlzworth HX4210 and the LTC7957-4. I have found that the performance of these devices is comparable with a 10MHz sinewave input with a PN noise floor below -160dBc/Hz. For offsets below 100Hz the LTC6957-4 is quieter and for offsets below 50Hz or so is lower than that of my measurement setup. I merely mounted the LTC evaluation board in a diecast metal box to reduce the effect of air currents on the LTC7957-4 and used an ultra low noise 3.3V power supply (Abracon) to minimise the power supply contribution to the measured phase noise. Achieving a phase noise performance equal to that on the datasheet was relatively easy. In fact for low offsets the phase noise is lower than that shown on the datasheet. I've also measured the phase noise of a comparator based sine to CMOS converter and its PN floor is about 20dB higher than that of the LTC6957. On completion of measurements PN plots for the various sine to square conveters tested will be added to the web pages I maintain on Didiers site.. Bruce ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Sine to square wave converter Phase Noise
Am 29.07.2014 um 13:38 schrieb David C. Partridge: Further to my previous post: I note that the LTC6957 is 3.3V CMOS rather than 5V. Do you have a recommmendation for a level converter to 5V logic? Would a 74AC04 using a 5V supply cope well enough in that role or is there a better solution? 74ACT04. Keeps the low input trip point that used to be common with TTL. http://www.fairchildsemi.com/ds/74/74ACT04.pdf regards, Gerhard ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Sine to square wave converter Phase Noise
However that will increase the signal transition times to significanly greater than the 300ps seen at the LTC6957 output. Bruce On July 29, 2014 at 7:54 AM Gerhard Hoffmann dk...@arcor.de wrote: Am 29.07.2014 um 13:38 schrieb David C. Partridge: Further to my previous post: I note that the LTC6957 is 3.3V CMOS rather than 5V. Do you have a recommmendation for a level converter to 5V logic? Would a 74AC04 using a 5V supply cope well enough in that role or is there a better solution? 74ACT04. Keeps the low input trip point that used to be common with TTL. http://www.fairchildsemi.com/ds/74/74ACT04.pdf regards, Gerhard ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Sine to square wave converter Phase Noise
I merely used the evaluation board which has a capacitively coupled input. The transformer coupled input circuit using a 1:4 (turns ratio) step up transformer as depicted on the datasheet should reduce the PN floor by at least 3dB for 10MHz input. Bruce On July 29, 2014 at 7:30 AM David C. Partridge david.partri...@perdrix.co.uk wrote: Bruce, Thank you, that was very useful information. If I ever do a re-spin of the divider I will likely use the LTC6957-3 or -4 instead of the ADCMP600 (I would need interest for at least 40 boards to be able to do this at anything like an economic price). Will you be putting the full details of the circuit you used up on Didier's web site? Did you use a transformer based circuit similar to Design Note 514 or just an input capacitor? Regards, David Partridge -Original Message- From: time-nuts-boun...@febo.com [mailto:time-nuts-boun...@febo.com] On Behalf Of br...@ko4bb.com Sent: 29 July 2014 12:00 To: Discussion of precise time and frequency measurement Subject: Re: [time-nuts] Sine to square wave converter Phase Noise Sorry, I meant the LTC6957-4 of course._SELA = l FILTA = L FILTB = H ie an input stage BW of 160MHz as recommended for a 10MHz + 10dBm input. Input for the HX4210 was +14dBm. Comparator was that used in David Partridge's programmable divider (includes the 74AC04 buffers). Bruce On July 29, 2014 at 5:37 AM Magnus Danielson mag...@rubidium.dyndns.org wrote: You alternate LTC7957-4 and LTC6957-4 and it is only the later that exists. The LTC6957-4 has a single-stage amplifier stage input, with somewhat programmable bandwidth. I assume you used the 50 MHz BW setting. Cheers, Magnus On 07/29/2014 10:55 AM, br...@ko4bb.com wrote: Recently I have been comparing the phase noise of the HOlzworth HX4210 and the LTC7957-4. I have found that the performance of these devices is comparable with a 10MHz sinewave input with a PN noise floor below -160dBc/Hz. For offsets below 100Hz the LTC6957-4 is quieter and for offsets below 50Hz or so is lower than that of my measurement setup. I merely mounted the LTC evaluation board in a diecast metal box to reduce the effect of air currents on the LTC7957-4 and used an ultra low noise 3.3V power supply (Abracon) to minimise the power supply contribution to the measured phase noise. Achieving a phase noise performance equal to that on the datasheet was relatively easy. In fact for low offsets the phase noise is lower than that shown on the datasheet. I've also measured the phase noise of a comparator based sine to CMOS converter and its PN floor is about 20dB higher than that of the LTC6957. On completion of measurements PN plots for the various sine to square conveters tested will be added to the web pages I maintain on Didiers site.. Bruce ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Sine to square wave converter Phase Noise
Am 29.07.2014 um 14:47 schrieb br...@ko4bb.com: However that will increase the signal transition times to significanly greater than the 300ps seen at the LTC6957 output. I had standard sot23/sc70 Fairchild single cmos gates with rise/fall times faster than 500ps. Gerhard ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Sine to square wave converter Phase Noise
One could possibly use another transistor pair in front of this one. For lower rate signals, it is a great finalizer solution to a step-wise increase in bandwidth and slew-rate. Cheers, Magnus On 07/29/2014 02:54 PM, br...@ko4bb.com wrote: I merely used the evaluation board which has a capacitively coupled input. The transformer coupled input circuit using a 1:4 (turns ratio) step up transformer as depicted on the datasheet should reduce the PN floor by at least 3dB for 10MHz input. Bruce On July 29, 2014 at 7:30 AM David C. Partridge david.partri...@perdrix.co.uk wrote: Bruce, Thank you, that was very useful information. If I ever do a re-spin of the divider I will likely use the LTC6957-3 or -4 instead of the ADCMP600 (I would need interest for at least 40 boards to be able to do this at anything like an economic price). Will you be putting the full details of the circuit you used up on Didier's web site? Did you use a transformer based circuit similar to Design Note 514 or just an input capacitor? Regards, David Partridge -Original Message- From: time-nuts-boun...@febo.com [mailto:time-nuts-boun...@febo.com] On Behalf Of br...@ko4bb.com Sent: 29 July 2014 12:00 To: Discussion of precise time and frequency measurement Subject: Re: [time-nuts] Sine to square wave converter Phase Noise Sorry, I meant the LTC6957-4 of course._SELA = l FILTA = L FILTB = H ie an input stage BW of 160MHz as recommended for a 10MHz + 10dBm input. Input for the HX4210 was +14dBm. Comparator was that used in David Partridge's programmable divider (includes the 74AC04 buffers). Bruce On July 29, 2014 at 5:37 AM Magnus Danielson mag...@rubidium.dyndns.org wrote: You alternate LTC7957-4 and LTC6957-4 and it is only the later that exists. The LTC6957-4 has a single-stage amplifier stage input, with somewhat programmable bandwidth. I assume you used the 50 MHz BW setting. Cheers, Magnus On 07/29/2014 10:55 AM, br...@ko4bb.com wrote: Recently I have been comparing the phase noise of the HOlzworth HX4210 and the LTC7957-4. I have found that the performance of these devices is comparable with a 10MHz sinewave input with a PN noise floor below -160dBc/Hz. For offsets below 100Hz the LTC6957-4 is quieter and for offsets below 50Hz or so is lower than that of my measurement setup. I merely mounted the LTC evaluation board in a diecast metal box to reduce the effect of air currents on the LTC7957-4 and used an ultra low noise 3.3V power supply (Abracon) to minimise the power supply contribution to the measured phase noise. Achieving a phase noise performance equal to that on the datasheet was relatively easy. In fact for low offsets the phase noise is lower than that shown on the datasheet. I've also measured the phase noise of a comparator based sine to CMOS converter and its PN floor is about 20dB higher than that of the LTC6957. On completion of measurements PN plots for the various sine to square conveters tested will be added to the web pages I maintain on Didiers site.. Bruce ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Sine to square wave converter Phase Noise
Hej Bruce, Many thanks for the clarification! This could be an alternate input stage to the TADD-2 too. Cheers, Magnus On 07/29/2014 12:59 PM, br...@ko4bb.com wrote: Sorry, I meant the LTC6957-4 of course._SELA = l FILTA = L FILTB = H ie an input stage BW of 160MHz as recommended for a 10MHz + 10dBm input. Input for the HX4210 was +14dBm. Comparator was that used in David Partridge's programmable divider (includes the 74AC04 buffers). Bruce On July 29, 2014 at 5:37 AM Magnus Danielson mag...@rubidium.dyndns.org wrote: You alternate LTC7957-4 and LTC6957-4 and it is only the later that exists. The LTC6957-4 has a single-stage amplifier stage input, with somewhat programmable bandwidth. I assume you used the 50 MHz BW setting. Cheers, Magnus On 07/29/2014 10:55 AM, br...@ko4bb.com wrote: Recently I have been comparing the phase noise of the HOlzworth HX4210 and the LTC7957-4. I have found that the performance of these devices is comparable with a 10MHz sinewave input with a PN noise floor below -160dBc/Hz. For offsets below 100Hz the LTC6957-4 is quieter and for offsets below 50Hz or so is lower than that of my measurement setup. I merely mounted the LTC evaluation board in a diecast metal box to reduce the effect of air currents on the LTC7957-4 and used an ultra low noise 3.3V power supply (Abracon) to minimise the power supply contribution to the measured phase noise. Achieving a phase noise performance equal to that on the datasheet was relatively easy. In fact for low offsets the phase noise is lower than that shown on the datasheet. I've also measured the phase noise of a comparator based sine to CMOS converter and its PN floor is about 20dB higher than that of the LTC6957. On completion of measurements PN plots for the various sine to square conveters tested will be added to the web pages I maintain on Didiers site.. Bruce ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Sine to square wave converter Phase Noise
Hi, What comparator did you use in your evaluation? An 74AC04, 74AC14 or other LT, AD, etc? Regards, Vasco Soares - Original Message - From: br...@ko4bb.com To: time-nuts@febo.com Sent: Tuesday, July 29, 2014 9:55 AM Subject: [time-nuts] Sine to square wave converter Phase Noise Recently I have been comparing the phase noise of the HOlzworth HX4210 and the LTC7957-4. I have found that the performance of these devices is comparable with a 10MHz sinewave input with a PN noise floor below -160dBc/Hz. For offsets below 100Hz the LTC6957-4 is quieter and for offsets below 50Hz or so is lower than that of my measurement setup. I merely mounted the LTC evaluation board in a diecast metal box to reduce the effect of air currents on the LTC7957-4 and used an ultra low noise 3.3V power supply (Abracon) to minimise the power supply contribution to the measured phase noise. Achieving a phase noise performance equal to that on the datasheet was relatively easy. In fact for low offsets the phase noise is lower than that shown on the datasheet. I've also measured the phase noise of a comparator based sine to CMOS converter and its PN floor is about 20dB higher than that of the LTC6957. On completion of measurements PN plots for the various sine to square conveters tested will be added to the web pages I maintain on Didiers site.. Bruce ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] Sine to square wave converter Phase Noise
Hi Put an L network in front of a biased NC7SZ04 running off of 5V and you should get a floor that’s ~ -170 dbc/Hz at 10 MHz. $0.34 each at quantity =1 at Mouser. The L network is set up to “what ever it takes” to give you 5V p-p on the gate input. The load termination on the L acts as the 50 ohm load for the oscillator. Bob On Jul 29, 2014, at 8:54 AM, br...@ko4bb.com wrote: I merely used the evaluation board which has a capacitively coupled input. The transformer coupled input circuit using a 1:4 (turns ratio) step up transformer as depicted on the datasheet should reduce the PN floor by at least 3dB for 10MHz input. Bruce On July 29, 2014 at 7:30 AM David C. Partridge david.partri...@perdrix.co.uk wrote: Bruce, Thank you, that was very useful information. If I ever do a re-spin of the divider I will likely use the LTC6957-3 or -4 instead of the ADCMP600 (I would need interest for at least 40 boards to be able to do this at anything like an economic price). Will you be putting the full details of the circuit you used up on Didier's web site? Did you use a transformer based circuit similar to Design Note 514 or just an input capacitor? Regards, David Partridge -Original Message- From: time-nuts-boun...@febo.com [mailto:time-nuts-boun...@febo.com] On Behalf Of br...@ko4bb.com Sent: 29 July 2014 12:00 To: Discussion of precise time and frequency measurement Subject: Re: [time-nuts] Sine to square wave converter Phase Noise Sorry, I meant the LTC6957-4 of course._SELA = l FILTA = L FILTB = H ie an input stage BW of 160MHz as recommended for a 10MHz + 10dBm input. Input for the HX4210 was +14dBm. Comparator was that used in David Partridge's programmable divider (includes the 74AC04 buffers). Bruce On July 29, 2014 at 5:37 AM Magnus Danielson mag...@rubidium.dyndns.org wrote: You alternate LTC7957-4 and LTC6957-4 and it is only the later that exists. The LTC6957-4 has a single-stage amplifier stage input, with somewhat programmable bandwidth. I assume you used the 50 MHz BW setting. Cheers, Magnus On 07/29/2014 10:55 AM, br...@ko4bb.com wrote: Recently I have been comparing the phase noise of the HOlzworth HX4210 and the LTC7957-4. I have found that the performance of these devices is comparable with a 10MHz sinewave input with a PN noise floor below -160dBc/Hz. For offsets below 100Hz the LTC6957-4 is quieter and for offsets below 50Hz or so is lower than that of my measurement setup. I merely mounted the LTC evaluation board in a diecast metal box to reduce the effect of air currents on the LTC7957-4 and used an ultra low noise 3.3V power supply (Abracon) to minimise the power supply contribution to the measured phase noise. Achieving a phase noise performance equal to that on the datasheet was relatively easy. In fact for low offsets the phase noise is lower than that shown on the datasheet. I've also measured the phase noise of a comparator based sine to CMOS converter and its PN floor is about 20dB higher than that of the LTC6957. On completion of measurements PN plots for the various sine to square conveters tested will be added to the web pages I maintain on Didiers site.. Bruce ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] sine to square wave converter
The Achilles heel of that biasing technique is that the emitter currents of the pair of pnp's is affected by the noise on the 20V supply. The 20V supply noise is only attenuated by a factor of 5 or so when both transistors have equal collector currents. I have both an LTC evaluation board for the LTC6957 and a Holzworth sine to CMOS converter as well as a Timepod so I could measure the phase noise of both of these for various 10MHz input signal levels. Bruce On July 10, 2014 at 8:21 PM Charles Steinmetz csteinm...@yandex.com wrote: Bruce wrote: Currently Linear Technology's sine to square wave devices with selectable filtering (LTC6957 series) are better in that they are a closer approximation to the ideal zero crossing detector. Failing that the next best is perhaps an AC coupled (both at input and between emitters) differential pair of 2N3906's or similar. My initial results with the LTC6957 did not produce lower phase noise at 10MHz than an optimized Wenzel two-PNP circuit (it may be possible to do better than my initial experiments with the 6957). Here is the circuit I use: Emacs! Using a 20v supply reduces the input feedthrough due to Q1's B-E capacitance, which tends to give the output square wave a sloping top. Using MPSH81s rather than 2N3906s helps with feedthrough, also, as well as reducing the rise and fall times (both about 2-4 nS with this circuit, depending on how hard it is driven, if it is built with proper attention to layout and stray capacitance). Some will insist that the LM329 is overkill, but the base bias can be a significant source (even the dominant source) of phase noise/jitter. The stability and low noise of the 329 improve performance materially -- even a TL431 or 1N829 is measurably inferior. An LM399 is somewhat better than the 329, but I have not found it necessary in practice. (Note that the pullup resistor is not shown -- 1.5k to 10k metal film from the 329 to +20v, not critical.) Some additional improvement can be achieved by using the PNP devices in an HFA3096 or HFA3128 array, but I have generally not seen the need for this in practic. As drawn, this circuit has lower residual PN than any 10MHz oscillator I have measured. Works best with input levels from 1 to 10Vpp (350mV to 3.5Vrms sine wave). There is a small duty cycle asymmetry (high longer than low), which depends on drive level. Using faster devices (such as HFA3096 or HFA3128) reduces the asymmetry. If this is a problem, a resistor can be added from the base of Q1 to ground to trim out the asymmetry if the input level is well controlled. Otherwise, the mean output voltage can be detected, compared to a reference, and used to adjust either base voltage with a servo loop. Best regards, Charles ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] sine to square wave converter
Bruce, please make those tests it will help me to how to proceed and probably many other time nuts. Bert Kehren In a message dated 7/11/2014 3:50:36 A.M. Eastern Daylight Time, br...@ko4bb.com writes: The Achilles heel of that biasing technique is that the emitter currents of the pair of pnp's is affected by the noise on the 20V supply. The 20V supply noise is only attenuated by a factor of 5 or so when both transistors have equal collector currents. I have both an LTC evaluation board for the LTC6957 and a Holzworth sine to CMOS converter as well as a Timepod so I could measure the phase noise of both of these for various 10MHz input signal levels. Bruce On July 10, 2014 at 8:21 PM Charles Steinmetz csteinm...@yandex.com wrote: Bruce wrote: Currently Linear Technology's sine to square wave devices with selectable filtering (LTC6957 series) are better in that they are a closer approximation to the ideal zero crossing detector. Failing that the next best is perhaps an AC coupled (both at input and between emitters) differential pair of 2N3906's or similar. My initial results with the LTC6957 did not produce lower phase noise at 10MHz than an optimized Wenzel two-PNP circuit (it may be possible to do better than my initial experiments with the 6957). Here is the circuit I use: Emacs! Using a 20v supply reduces the input feedthrough due to Q1's B-E capacitance, which tends to give the output square wave a sloping top. Using MPSH81s rather than 2N3906s helps with feedthrough, also, as well as reducing the rise and fall times (both about 2-4 nS with this circuit, depending on how hard it is driven, if it is built with proper attention to layout and stray capacitance). Some will insist that the LM329 is overkill, but the base bias can be a significant source (even the dominant source) of phase noise/jitter. The stability and low noise of the 329 improve performance materially -- even a TL431 or 1N829 is measurably inferior. An LM399 is somewhat better than the 329, but I have not found it necessary in practice. (Note that the pullup resistor is not shown -- 1.5k to 10k metal film from the 329 to +20v, not critical.) Some additional improvement can be achieved by using the PNP devices in an HFA3096 or HFA3128 array, but I have generally not seen the need for this in practic. As drawn, this circuit has lower residual PN than any 10MHz oscillator I have measured. Works best with input levels from 1 to 10Vpp (350mV to 3.5Vrms sine wave). There is a small duty cycle asymmetry (high longer than low), which depends on drive level. Using faster devices (such as HFA3096 or HFA3128) reduces the asymmetry. If this is a problem, a resistor can be added from the base of Q1 to ground to trim out the asymmetry if the input level is well controlled. Otherwise, the mean output voltage can be detected, compared to a reference, and used to adjust either base voltage with a servo loop. Best regards, Charles ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] sine to square wave converter
Bruce wrote: The Achilles heel of that biasing technique is that the emitter currents of the pair of pnp's is affected by the noise on the 20V supply. Yes, you do need a quiet supply -- I suppose I should have mentioned that. But sufficiently quiet supplies are not hard to design, they're just more complex than the has to be done with just one ten-cent part crowd wants (don't expect to use integrated regulators of any description). Bruce -- do you know the topology of the Holzworth converter? Best regards, Charles ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
[time-nuts] sine to square wave converter
As part of the FE 405 B project a separate output circuit is in the works. The universal controller and auxiliary board are the same as used in the FE5680A GPSDO but because of the very low ADEV a separate circuit board that divides by three and has also two ground isolated transformer outputs is in the works.The question is what is the best sine to square wave converter with the lowest ADEV contribution. I am looking at Bruce's circuit using the ADCMP600. Any other ideas? Thanks Bert Kehren ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] sine to square wave converter
On July 10, 2014 at 10:55 AM ewkeh...@aol.com wrote: As part of the FE 405 B project a separate output circuit is in the works. The universal controller and auxiliary board are the same as used in the FE5680A GPSDO but because of the very low ADEV a separate circuit board that divides by three and has also two ground isolated transformer outputs is in the works.The question is what is the best sine to square wave converter with the lowest ADEV contribution. I am looking at Bruce's circuit using the ADCMP600. Any other ideas? Thanks Bert Kehren ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. Although CERN likes that circuit for its ruggedness and versatility its a little noisier than some alternative circuits. Currently Linear Technology's sine to square wave devices with selectable filtering (LTC6957 series) are better in that they are a closer approximation to the ideal zero crossing detector. Failing that the next best is perhaps an AC coupled (both at input and between emitters) differential pair of 2N3906's or similar. Bruce ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] sine to square wave converter
look Charles Wenzel's waveform conversion: http://www.wenzel.com/documents/waveform.html 73 Alex On 7/10/2014 11:40 AM, br...@ko4bb.com wrote: On July 10, 2014 at 10:55 AM ewkeh...@aol.com wrote: As part of the FE 405 B project a separate output circuit is in the works. The universal controller and auxiliary board are the same as used in the FE5680A GPSDO but because of the very low ADEV a separate circuit board that divides by three and has also two ground isolated transformer outputs is in the works.The question is what is the best sine to square wave converter with the lowest ADEV contribution. I am looking at Bruce's circuit using the ADCMP600. Any other ideas? Thanks Bert Kehren ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. Although CERN likes that circuit for its ruggedness and versatility its a little noisier than some alternative circuits. Currently Linear Technology's sine to square wave devices with selectable filtering (LTC6957 series) are better in that they are a closer approximation to the ideal zero crossing detector. Failing that the next best is perhaps an AC coupled (both at input and between emitters) differential pair of 2N3906's or similar. Bruce ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] sine to square wave converter
Bert, On 07/10/2014 04:55 PM, ewkeh...@aol.com wrote: As part of the FE 405 B project a separate output circuit is in the works. The universal controller and auxiliary board are the same as used in the FE5680A GPSDO but because of the very low ADEV a separate circuit board that divides by three and has also two ground isolated transformer outputs is in the works.The question is what is the best sine to square wave converter with the lowest ADEV contribution. I am looking at Bruce's circuit using the ADCMP600. Any other ideas? Do look at the Wenzel clockshaper [1], look at the TADD-2 [2] schematic. It's a PNP long-tail pair. The strategy is to provide modest gain. A known strategy to reduce 1/f noise and to some degree thermal differences is naturally feedback, as you will find in the NIST papers. Once you have the slew-rate up, going in for the kill with a straight comparator should give you all the nice output slew-rate you can wish for. Thus, this is not all that different to the mixer-setup you have done. I have modified my TADD-2 such that one of the output channels is fed from the input circuit, and this provides me with a squared up version. For a counter such as DTS-2070C, the difference is significant, which helps to show the potential of this simple design. I think the basic approach can be improved, and how far one has to go depends on how clean source you have. You end up with interesting measurement problems. An indirect way to measure the goodness of a squarer is to insert some known sine disturbance at say 30 or 40 dB below the signal. A straight comparator won't work very well. Be careful with selectivity of LC, as it is a nice way to become temperature dependent, so low-Q solutions is needed. Cheers, Magnus [1] http://www.wenzel.com/documents/waveform.html [2] https://www.tapr.org/kits_tadd-2.html ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] sine to square wave converter
Thank you Alex I have that info I am concerned about rise time Bert Kehren Sent from Samsung tabletAlexander Pummer alex...@ieee.org wrote: look Charles Wenzel's waveform conversion: http://www.wenzel.com/documents/waveform.html 73 Alex On 7/10/2014 11:40 AM, br...@ko4bb.com wrote: On July 10, 2014 at 10:55 AM ewkeh...@aol.com wrote: As part of the FE 405 B project a separate output circuit is in the works. The universal controller and auxiliary board are the same as used in the FE5680A GPSDO but because of the very low ADEV a separate circuit board that divides by three and has also two ground isolated transformer outputs is in the works.The question is what is the best sine to square wave converter with the lowest ADEV contribution. I am looking at Bruce's circuit using the ADCMP600. Any other ideas? Thanks Bert Kehren ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. Although CERN likes that circuit for its ruggedness and versatility its a little noisier than some alternative circuits. Currently Linear Technology's sine to square wave devices with selectable filtering (LTC6957 series) are better in that they are a closer approximation to the ideal zero crossing detector. Failing that the next best is perhaps an AC coupled (both at input and between emitters) differential pair of 2N3906's or similar. Bruce ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] sine to square wave converter
Magnus Thank you for your recommendation I use Wenzel extensively as a matter of fact I just completed in the last three days two boards that have Wenzel on it and in my projects I can count 14 boards. Rise and fall time is my concern but I am open to suggestions that is why I turned to the list looking for the best. Thanks again Bert Kehren In a message dated 7/10/2014 3:09:52 P.M. Eastern Daylight Time, mag...@rubidium.dyndns.org writes: Bert, On 07/10/2014 04:55 PM, ewkeh...@aol.com wrote: As part of the FE 405 B project a separate output circuit is in the works. The universal controller and auxiliary board are the same as used in the FE5680A GPSDO but because of the very low ADEV a separate circuit board that divides by three and has also two ground isolated transformer outputs is in the works.The question is what is the best sine to square wave converter with the lowest ADEV contribution. I am looking at Bruce's circuit using the ADCMP600. Any other ideas? Do look at the Wenzel clockshaper [1], look at the TADD-2 [2] schematic. It's a PNP long-tail pair. The strategy is to provide modest gain. A known strategy to reduce 1/f noise and to some degree thermal differences is naturally feedback, as you will find in the NIST papers. Once you have the slew-rate up, going in for the kill with a straight comparator should give you all the nice output slew-rate you can wish for. Thus, this is not all that different to the mixer-setup you have done. I have modified my TADD-2 such that one of the output channels is fed from the input circuit, and this provides me with a squared up version. For a counter such as DTS-2070C, the difference is significant, which helps to show the potential of this simple design. I think the basic approach can be improved, and how far one has to go depends on how clean source you have. You end up with interesting measurement problems. An indirect way to measure the goodness of a squarer is to insert some known sine disturbance at say 30 or 40 dB below the signal. A straight comparator won't work very well. Be careful with selectivity of LC, as it is a nice way to become temperature dependent, so low-Q solutions is needed. Cheers, Magnus [1] http://www.wenzel.com/documents/waveform.html [2] https://www.tapr.org/kits_tadd-2.html ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to ht tps://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] sine to square wave converter
Bert, OK, good that you are familiar with it, it was not obvious in that message. If you consider it as the first stage, and that you then can put another (faster) stage after it until you go for comparator. It's just the same thing as the multistage for beat-note, but you run at a higher frequency. That way you should increase your slew-rate step-wise. Cheers, Magnus On 07/10/2014 09:37 PM, ewkeh...@aol.com wrote: Magnus Thank you for your recommendation I use Wenzel extensively as a matter of fact I just completed in the last three days two boards that have Wenzel on it and in my projects I can count 14 boards. Rise and fall time is my concern but I am open to suggestions that is why I turned to the list looking for the best. Thanks again Bert Kehren In a message dated 7/10/2014 3:09:52 P.M. Eastern Daylight Time, mag...@rubidium.dyndns.org writes: Bert, On 07/10/2014 04:55 PM, ewkeh...@aol.com wrote: As part of the FE 405 B project a separate output circuit is in the works. The universal controller and auxiliary board are the same as used in the FE5680A GPSDO but because of the very low ADEV a separate circuit board that divides by three and has also two ground isolated transformer outputs is in the works.The question is what is the best sine to square wave converter with the lowest ADEV contribution. I am looking at Bruce's circuit using the ADCMP600. Any other ideas? Do look at the Wenzel clockshaper [1], look at the TADD-2 [2] schematic. It's a PNP long-tail pair. The strategy is to provide modest gain. A known strategy to reduce 1/f noise and to some degree thermal differences is naturally feedback, as you will find in the NIST papers. Once you have the slew-rate up, going in for the kill with a straight comparator should give you all the nice output slew-rate you can wish for. Thus, this is not all that different to the mixer-setup you have done. I have modified my TADD-2 such that one of the output channels is fed from the input circuit, and this provides me with a squared up version. For a counter such as DTS-2070C, the difference is significant, which helps to show the potential of this simple design. I think the basic approach can be improved, and how far one has to go depends on how clean source you have. You end up with interesting measurement problems. An indirect way to measure the goodness of a squarer is to insert some known sine disturbance at say 30 or 40 dB below the signal. A straight comparator won't work very well. Be careful with selectivity of LC, as it is a nice way to become temperature dependent, so low-Q solutions is needed. Cheers, Magnus [1] http://www.wenzel.com/documents/waveform.html [2] https://www.tapr.org/kits_tadd-2.html ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to ht tps://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] sine to square wave converter
Will the added stage negatively effect ADEV? Bert In a message dated 7/10/2014 3:48:44 P.M. Eastern Daylight Time, mag...@rubidium.dyndns.org writes: Bert, OK, good that you are familiar with it, it was not obvious in that message. If you consider it as the first stage, and that you then can put another (faster) stage after it until you go for comparator. It's just the same thing as the multistage for beat-note, but you run at a higher frequency. That way you should increase your slew-rate step-wise. Cheers, Magnus On 07/10/2014 09:37 PM, ewkeh...@aol.com wrote: Magnus Thank you for your recommendation I use Wenzel extensively as a matter of fact I just completed in the last three days two boards that have Wenzel on it and in my projects I can count 14 boards. Rise and fall time is my concern but I am open to suggestions that is why I turned to the list looking for the best. Thanks againBert Kehren In a message dated 7/10/2014 3:09:52 P.M. Eastern Daylight Time, mag...@rubidium.dyndns.org writes: Bert, On 07/10/2014 04:55 PM, ewkeh...@aol.com wrote: As part of the FE 405 B project a separate output circuit is in the works. The universal controller and auxiliary board are the same as used in the FE5680A GPSDO but because of the very low ADEV a separate circuit board that divides by three and has also two ground isolated transformer outputs is in the works.The question is what is the best sine to square wave converter with the lowest ADEV contribution. I am looking at Bruce's circuit using the ADCMP600. Any other ideas? Do look at the Wenzel clockshaper [1], look at the TADD-2 [2] schematic. It's a PNP long-tail pair. The strategy is to provide modest gain. A known strategy to reduce 1/f noise and to some degree thermal differences is naturally feedback, as you will find in the NIST papers. Once you have the slew-rate up, going in for the kill with a straight comparator should give you all the nice output slew-rate you can wish for. Thus, this is not all that different to the mixer-setup you have done. I have modified my TADD-2 such that one of the output channels is fed from the input circuit, and this provides me with a squared up version. For a counter such as DTS-2070C, the difference is significant, which helps to show the potential of this simple design. I think the basic approach can be improved, and how far one has to go depends on how clean source you have. You end up with interesting measurement problems. An indirect way to measure the goodness of a squarer is to insert some known sine disturbance at say 30 or 40 dB below the signal. A straight comparator won't work very well. Be careful with selectivity of LC, as it is a nice way to become temperature dependent, so low-Q solutions is needed. Cheers, Magnus [1] http://www.wenzel.com/documents/waveform.html [2] https://www.tapr.org/kits_tadd-2.html ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to ht tps://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] sine to square wave converter
How much of an improvement does the LTC6957 give over, say, a simple 74AC gate? Actually, is there a list of different techniques and the performance of each? I know that the LPRO Integration manual gives some good info for phase noise, but I'm not aware of a similar document for slew rate (AKA rise time). Ed On 7/10/2014 12:40 PM, br...@ko4bb.com wrote: On July 10, 2014 at 10:55 AM ewkeh...@aol.com wrote: As part of the FE 405 B project a separate output circuit is in the works. The universal controller and auxiliary board are the same as used in the FE5680A GPSDO but because of the very low ADEV a separate circuit board that divides by three and has also two ground isolated transformer outputs is in the works.The question is what is the best sine to square wave converter with the lowest ADEV contribution. I am looking at Bruce's circuit using the ADCMP600. Any other ideas? Thanks Bert Kehren ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. Although CERN likes that circuit for its ruggedness and versatility its a little noisier than some alternative circuits. Currently Linear Technology's sine to square wave devices with selectable filtering (LTC6957 series) are better in that they are a closer approximation to the ideal zero crossing detector. Failing that the next best is perhaps an AC coupled (both at input and between emitters) differential pair of 2N3906's or similar. Bruce ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] sine to square wave converter
Bert, Hard to predict without looking at the details. The main battle for a sine-to-square is to fight white and flicker phase noise as well as additive repetitive signals such as 50 Hz power noise creeping into the jitter. At the same time we try to get the conversion more long-term stable, so that temperature and voltage changes has less impact on the comparator point as it will convert voltage shifts into phase-shifts. However, as any active device, it too will add noise and phase-shifts. I think that this is a field where you need to experiment and learn, and I try to show some alternative routes and some known traps which may solve one problem but cause another. So, properly built, they should contribute if this is where you are having troubles, if it's not, they will add to long term ADEV for sure. Cheers, Magnus On 07/10/2014 09:51 PM, ewkeh...@aol.com wrote: Will the added stage negatively effect ADEV? Bert In a message dated 7/10/2014 3:48:44 P.M. Eastern Daylight Time, mag...@rubidium.dyndns.org writes: Bert, OK, good that you are familiar with it, it was not obvious in that message. If you consider it as the first stage, and that you then can put another (faster) stage after it until you go for comparator. It's just the same thing as the multistage for beat-note, but you run at a higher frequency. That way you should increase your slew-rate step-wise. Cheers, Magnus On 07/10/2014 09:37 PM, ewkeh...@aol.com wrote: Magnus Thank you for your recommendation I use Wenzel extensively as a matter of fact I just completed in the last three days two boards that have Wenzel on it and in my projects I can count 14 boards. Rise and fall time is my concern but I am open to suggestions that is why I turned to the list looking for the best. Thanks againBert Kehren In a message dated 7/10/2014 3:09:52 P.M. Eastern Daylight Time, mag...@rubidium.dyndns.org writes: Bert, On 07/10/2014 04:55 PM, ewkeh...@aol.com wrote: As part of the FE 405 B project a separate output circuit is in the works. The universal controller and auxiliary board are the same as used in the FE5680A GPSDO but because of the very low ADEV a separate circuit board that divides by three and has also two ground isolated transformer outputs is in the works.The question is what is the best sine to square wave converter with the lowest ADEV contribution. I am looking at Bruce's circuit using the ADCMP600. Any other ideas? Do look at the Wenzel clockshaper [1], look at the TADD-2 [2] schematic. It's a PNP long-tail pair. The strategy is to provide modest gain. A known strategy to reduce 1/f noise and to some degree thermal differences is naturally feedback, as you will find in the NIST papers. Once you have the slew-rate up, going in for the kill with a straight comparator should give you all the nice output slew-rate you can wish for. Thus, this is not all that different to the mixer-setup you have done. I have modified my TADD-2 such that one of the output channels is fed from the input circuit, and this provides me with a squared up version. For a counter such as DTS-2070C, the difference is significant, which helps to show the potential of this simple design. I think the basic approach can be improved, and how far one has to go depends on how clean source you have. You end up with interesting measurement problems. An indirect way to measure the goodness of a squarer is to insert some known sine disturbance at say 30 or 40 dB below the signal. A straight comparator won't work very well. Be careful with selectivity of LC, as it is a nice way to become temperature dependent, so low-Q solutions is needed. Cheers, Magnus [1] http://www.wenzel.com/documents/waveform.html [2] https://www.tapr.org/kits_tadd-2.html ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to ht tps://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there. ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
Re: [time-nuts] sine to square wave converter
The schematic of the converter I attached to my previous message did not make it through in my copy of the message, so it may be missing from other people's copies, as well. If anyone wants it and did not receive it, it is available at: http://www.febo.com/pipermail/time-nuts/attachments/20140710/9fb493f1/attachment-0001.jpg Best regards, Charles Bruce wrote: Currently Linear Technology's sine to square wave devices with selectable filtering (LTC6957 series) are better in that they are a closer approximation to the ideal zero crossing detector. Failing that the next best is perhaps an AC coupled (both at input and between emitters) differential pair of 2N3906's or similar. My initial results with the LTC6957 did not produce lower phase noise at 10MHz than an optimized Wenzel two-PNP circuit (it may be possible to do better than my initial experiments with the 6957). Here is the circuit I use: Emacs! Using a 20v supply reduces the input feedthrough due to Q1's B-E capacitance, which tends to give the output square wave a sloping top. Using MPSH81s rather than 2N3906s helps with feedthrough, also, as well as reducing the rise and fall times (both about 2-4 nS with this circuit, depending on how hard it is driven, if it is built with proper attention to layout and stray capacitance). Some will insist that the LM329 is overkill, but the base bias can be a significant source (even the dominant source) of phase noise/jitter. The stability and low noise of the 329 improve performance materially -- even a TL431 or 1N829 is measurably inferior. An LM399 is somewhat better than the 329, but I have not found it necessary in practice. (Note that the pullup resistor is not shown -- 1.5k to 10k metal film from the 329 to +20v, not critical.) Some additional improvement can be achieved by using the PNP devices in an HFA3096 or HFA3128 array, but I have generally not seen the need for this in practic. As drawn, this circuit has lower residual PN than any 10MHz oscillator I have measured. Works best with input levels from 1 to 10Vpp (350mV to 3.5Vrms sine wave). There is a small duty cycle asymmetry (high longer than low), which depends on drive level. Using faster devices (such as HFA3096 or HFA3128) reduces the asymmetry. If this is a problem, a resistor can be added from the base of Q1 to ground to trim out the asymmetry if the input level is well controlled. Otherwise, the mean output voltage can be detected, compared to a reference, and used to adjust either base voltage with a servo loop. Best regards, Charles ___ time-nuts mailing list -- time-nuts@febo.com To unsubscribe, go to https://www.febo.com/cgi-bin/mailman/listinfo/time-nuts and follow the instructions there.
