Re: [Drakelist] R-4B audio problem traced to low beta Q7 (2N3394)
On Sun, 17 Apr 2011 19:33:16 -0700, Richard Knoppow wrote: > General Radio also made a good low distortion audio >genrator and harmonic distortion meter. GR is nice stuff. I spotted a GR decade resistor box in the pile that I forgot I had when I was looking for the Leader distortion analyzer. It came in the house with me too. I picked up one of those retro looking GR sound level meters at an estate sale a couple of years ago, dirt cheap. I had seen them before, so I knew immediately what it was. You probably know it, the late 50s/early 60s hand-held model that used low voltage pencil tubes. All it needs are some brass accents to be steampunk. Other than some battery corrosion in the bottom, it is as pristine inside as something of that age can be. I rigged some power to it and it works FB, and I suspect it is still fairly close in calibration. It's just too cool not to keep and show off proudly. 73 -Jim -- Ham Radio NU0C Lincoln, Nebraska, U.S.S.A. TR7/RV7/R7A/L7, TR6/RV6, T4XC/R4C/L4B, NCL2000, SB104A, R390A, GT550A/RV550A, HyGain 3750, IBM PS/2 - all vintage, all the time! "Give a man a URL, and he will learn for an hour; teach him to Google, and he will learn for a lifetime." HyGain 3750 User's Group - http://groups.yahoo.com/group/HyGain_3750/ http://incolor.inetnebr.com/jshorney http://www.nebraskaghosts.org ___ Drakelist mailing list Drakelist@zerobeat.net http://mailman.zerobeat.net/mailman/listinfo/drakelist
Re: [Drakelist] R-4B audio problem traced to low beta Q7 (2N3394)
- Original Message - From: "Curt Nixon" To: "Jim Shorney" Cc: Sent: Sunday, April 17, 2011 5:13 PM Subject: Re: [Drakelist] R-4B audio problem traced to low beta Q7 (2N3394) HP had some mighty fine audio distortion analyzers and reference sine equipment in the 60's--we used it on the McIntosh stuff a lot. 600 series if I recall...same size as my RF gen. 19" rack mount..50 Lbs Curt General Radio also made a good low distortion audio genrator and harmonic distortion meter. Probably most of them were sold for broadcast station use in doing proof of performance testing but they were useful for production testing also. The generator had residual distortion on the order of 0.1%, high by modern standards but certainly low enough for routine testing of single ended tube amplifiers. I think the residual of the meter was lower. This is mid-1950's stuff. By the mid 1960s -hp- had much better generators and analysers available. -- Richard Knoppow Los Angeles WB6KBL dickb...@ix.netcom.com ___ Drakelist mailing list Drakelist@zerobeat.net http://mailman.zerobeat.net/mailman/listinfo/drakelist
Re: [Drakelist] R-4B audio problem traced to low beta Q7 (2N3394)
On Sun, 17 Apr 2011 20:13:18 -0400, Curt Nixon wrote: >HP had some mighty fine audio distortion analyzers and reference sine >equipment in the 60's--we used it on the McIntosh stuff a lot. > >600 series if I recall...same size as my RF gen. 19" rack mount..50 Lbs As I recall, mine are 300-series. More like 70s or 80s, I would guess. Still not light or small. 73 -Jim -- Ham Radio NU0C Lincoln, Nebraska, U.S.S.A. TR7/RV7/R7A/L7, TR6/RV6, T4XC/R4C/L4B, NCL2000, SB104A, R390A, GT550A/RV550A, HyGain 3750, IBM PS/2 - all vintage, all the time! "Give a man a URL, and he will learn for an hour; teach him to Google, and he will learn for a lifetime." HyGain 3750 User's Group - http://groups.yahoo.com/group/HyGain_3750/ http://incolor.inetnebr.com/jshorney http://www.nebraskaghosts.org ___ Drakelist mailing list Drakelist@zerobeat.net http://mailman.zerobeat.net/mailman/listinfo/drakelist
Re: [Drakelist] R-4B audio problem traced to low beta Q7 (2N3394)
HP had some mighty fine audio distortion analyzers and reference sine equipment in the 60's--we used it on the McIntosh stuff a lot. 600 series if I recall...same size as my RF gen. 19" rack mount..50 Lbs Curt Jim Shorney wrote: On Sun, 17 Apr 2011 13:07:35 -0700, Dennis Monticelli wrote: That would be an interesting study, going stage by stage through the audio channel of a 4 or 7 series radio. I'll bet that when the R-4B was designed, the Drake lab didn't have such a piece of equipment and probably went by ear or visuals on a scope. I've got a couple of HP and a Leader. The Leader gets the nod at the moment because it has a much smaller footprint. I serously need to declutter my workbench... 73 -Jim -- Ham Radio NU0C Lincoln, Nebraska, U.S.S.A. TR7/RV7/R7A/L7, TR6/RV6, T4XC/R4C/L4B, NCL2000, SB104A, R390A, GT550A/RV550A, HyGain 3750, IBM PS/2 - all vintage, all the time! "Give a man a URL, and he will learn for an hour; teach him to Google, and he will learn for a lifetime." HyGain 3750 User's Group - http://groups.yahoo.com/group/HyGain_3750/ http://incolor.inetnebr.com/jshorney http://www.nebraskaghosts.org ___ Drakelist mailing list Drakelist@zerobeat.net http://mailman.zerobeat.net/mailman/listinfo/drakelist ___ Drakelist mailing list Drakelist@zerobeat.net http://mailman.zerobeat.net/mailman/listinfo/drakelist
Re: [Drakelist] R-4B audio problem traced to low beta Q7 (2N3394)
OHHH Yeah...:) Curt Jim Shorney wrote: On Sun, 17 Apr 2011 14:23:02 -0700, Richard Knoppow wrote: The use of ceramic caps in audio circuits is somewhat controversial. Thanks for the info, Richard. There's been a good discussion of this on the R-390 list recently. I may have to go back and re-read some of it. 73 -Jim -- Ham Radio NU0C Lincoln, Nebraska, U.S.S.A. TR7/RV7/R7A/L7, TR6/RV6, T4XC/R4C/L4B, NCL2000, SB104A, R390A, GT550A/RV550A, HyGain 3750, IBM PS/2 - all vintage, all the time! "Give a man a URL, and he will learn for an hour; teach him to Google, and he will learn for a lifetime." HyGain 3750 User's Group - http://groups.yahoo.com/group/HyGain_3750/ http://incolor.inetnebr.com/jshorney http://www.nebraskaghosts.org ___ Drakelist mailing list Drakelist@zerobeat.net http://mailman.zerobeat.net/mailman/listinfo/drakelist ___ Drakelist mailing list Drakelist@zerobeat.net http://mailman.zerobeat.net/mailman/listinfo/drakelist
Re: [Drakelist] R-4B audio problem traced to low beta Q7 (2N3394)
On Sun, 17 Apr 2011 15:27:13 -0700, Dennis Monticelli wrote: >As I understand it, the only true concern one should have about the ceramics >is the piezo effect of the high K material. This makes the circuit >microphonic so unwanted acoustic feedback from the PA can take place. I agree >that voltage coeficient is not a significant effect unless you're trying to >build high accuracy stuff. In the world of integrated circuits of which I'm >familiar, the votlage coefficient only rears its ugly head beyond the 12b >level. At 16b it's a real concern, but we're talking very high linearity here. > Here's the stuff I just sent Richard off-list: FWIW, here's the most relevant portion of the R-390 discussion: --- Date: Thu, 31 Mar 2011 00:05:53 -0400 From: 2002tii Subject: Re: [R-390] Orange Drop vs ceramic disc There have been many tests, back to at least the '40s (long before small surfacemountceramics were a gleam in anyone's eye), documenting this behavior of ceramic caps. Bob Pease published a chart in EDN in the early '80s comparing the dielectric absorption-related distortion of various types of capacitors that showed measured distortion of 1% or more at low audio frequencies for ceramics (though NP0 caps were much better). (Note that this is a different mechanism than the voltage coefficient of capacitance, and that both mechanisms cause distortion independently.) We comprehensively tested all kinds of capacitors in audio coupling circuits in the '80s, and I assure you that the distortion from ceramic caps is clearly audible. Yes, we used a high-resolution audio system, not a communications radio -- but different types of distortion are heard more or less independently (that is, one type -- for example, even-order harmonic distortion, which dominates tube communications radio distortion -- does not mask another type -- for example, high-order intermodulation distortion or digital quantization errors, even when it is present at much higher levels), so the distortion due to ceramic coupling caps may very well be audible even in a circuit with 10% even-order harmonic distortion. We also used signal cancellation techniques to listen to the distortion products alone, and the distortion products of ceramic caps are extremely uglysounding (high-order, non harmonically-related -- easy to spot at very low levels). Since it is no effort whatsoever to avoid ceramic caps in coupling applications, there is simply no reason to use ceramics in those applications (indeed, it is simply good engineering practice to avoid possible ill effects when there is little or no cost to do so). --- Date: Thu, 31 Mar 2011 10:29:55 -0400 From: "Shoppa, Tim" Subject: Re: [R-390] Orange Drop vs ceramic disc Please keep in mind that when it comes to ceramic caps, the dielectric materials available vary widely in their characteristics. C0G or NP0 ceramic caps could well be golden in audio applications but don't have enough uF per package to be used in most situations. X7R ceramic caps are probably good enough for not-hi-fi applications. Indeed lots of low-end consumer tube stuff from the 60's used ceramic audio coupling caps. Y5V caps literally sound like crap in audio coupling circuits. The latest SMD ceramic caps and their MLCC leaded cousins, almost certainly outperform any 50 year old NOS ceramics we have lying about, every which way from Sunday. I look at the high-end microwave SMT ceramic caps (actually the highest end ones are glass) and they beat the pants off of any leaded component. --- Also, I found the Bob Pease article on the web: http://www.national.com/rap/Application/0,1570,28,00.html Then, there's this, which doesn't seem to be downloading very well at the moment: http://www.designers-guide.org/Modeling/da.pdf It all could very well be moot for communications radios, but OTOH, distortion does tend to add up. -- Ham Radio NU0C Lincoln, Nebraska, U.S.S.A. TR7/RV7/R7A/L7, TR6/RV6, T4XC/R4C/L4B, NCL2000, SB104A, R390A, GT550A/RV550A, HyGain 3750, IBM PS/2 - all vintage, all the time! "Give a man a URL, and he will learn for an hour; teach him to Google, and he will learn for a lifetime." HyGain 3750 User's Group - http://groups.yahoo.com/group/HyGain_3750/ http://incolor.inetnebr.com/jshorney http://www.nebraskaghosts.org ___ Drakelist mailing list Drakelist@zerobeat.net http://mailman.zerobeat.net/mailman/listinfo/drakelist
Re: [Drakelist] R-4B audio problem traced to low beta Q7 (2N3394)
As I understand it, the only true concern one should have about the ceramics is the piezo effect of the high K material. This makes the circuit microphonic so unwanted acoustic feedback from the PA can take place. I agree that voltage coeficient is not a significant effect unless you're trying to build high accuracy stuff. In the world of integrated circuits of which I'm familiar, the votlage coefficient only rears its ugly head beyond the 12b level. At 16b it's a real concern, but we're talking very high linearity here. Dennis AE6C On Sun, Apr 17, 2011 at 2:23 PM, Richard Knoppow <1oldle...@ix.netcom.com>wrote: > > - Original Message - From: "Jim Shorney" > To: > Sent: Sunday, April 17, 2011 12:59 PM > Subject: Re: [Drakelist] R-4B audio problem traced to low beta Q7 (2N3394) > > > > On Sun, 17 Apr 2011 15:13:23 -0400 (EDT), kc9...@aol.com wrote: >> >> Very good explanation...I understoof about 30% of it...but then I am a >>> tech., not an engineer. Hi, Hi. >>> >> >> What made this tough is that Q7 was kind of working...so I was >>> checking, replacing everything BUT Q7. >>> >> >> My experience as a tech in situations like this has taught me to suspect >> capacitors first, transistors next. I've seen transistors fail in some >> strange >> ways over the years. :) >> >> I've recently learned that disc ceramic caps are not ideal for audio >> coupling. >> You guys have got me thinking about distortion and such. I'm thinking >> about >> putting the TR-7 on the bench and trying out that nifty surplus distortion >> analyzer that's languising out in the garage. >> >> 73 >> >> -Jim >> >The use of ceramic caps in audio circuits is somewhat controversial. > Ceramics of some types, mainly high-K types, do tend to have a variation of > capacitance with voltage and some audio folks think this contributes some > distortion. However, in most coupling circuits there is a steady DC voltage > on the cap that exceeds the AC voltage. Also, I have never seen any rigorous > research on this showing that a real difference exists. High-K ceramics do > have high temperature coeficient of capacitance but low-K ones can have zero > temperature variation (NPO types). There can also be a small variation in > effective capacitance due to interfacial polarization (don't ask) but mica > caps have the same problem. This is of concern only were a cap is used for > precision measurement at low frequencies. Paper and plastic film caps have > virtually none of this effect. I doubt if it has any significance for normal > audio use. >Ceramic caps have the virtues of small sizes, availability in high > voltages, and great reliability and long life. > > > -- > Richard Knoppow > Los Angeles > WB6KBL > dickb...@ix.netcom.com > > > > ___ > Drakelist mailing list > Drakelist@zerobeat.net > http://mailman.zerobeat.net/mailman/listinfo/drakelist > ___ Drakelist mailing list Drakelist@zerobeat.net http://mailman.zerobeat.net/mailman/listinfo/drakelist
Re: [Drakelist] R-4B audio problem traced to low beta Q7 (2N3394)
On Sun, 17 Apr 2011 14:23:02 -0700, Richard Knoppow wrote: > The use of ceramic caps in audio circuits is somewhat >controversial. Thanks for the info, Richard. There's been a good discussion of this on the R-390 list recently. I may have to go back and re-read some of it. 73 -Jim -- Ham Radio NU0C Lincoln, Nebraska, U.S.S.A. TR7/RV7/R7A/L7, TR6/RV6, T4XC/R4C/L4B, NCL2000, SB104A, R390A, GT550A/RV550A, HyGain 3750, IBM PS/2 - all vintage, all the time! "Give a man a URL, and he will learn for an hour; teach him to Google, and he will learn for a lifetime." HyGain 3750 User's Group - http://groups.yahoo.com/group/HyGain_3750/ http://incolor.inetnebr.com/jshorney http://www.nebraskaghosts.org ___ Drakelist mailing list Drakelist@zerobeat.net http://mailman.zerobeat.net/mailman/listinfo/drakelist
Re: [Drakelist] R-4B audio problem traced to low beta Q7 (2N3394)
- Original Message - From: "Jim Shorney" To: Sent: Sunday, April 17, 2011 12:59 PM Subject: Re: [Drakelist] R-4B audio problem traced to low beta Q7 (2N3394) On Sun, 17 Apr 2011 15:13:23 -0400 (EDT), kc9...@aol.com wrote: Very good explanation...I understoof about 30% of it...but then I am a tech., not an engineer. Hi, Hi. What made this tough is that Q7 was kind of working...so I was checking, replacing everything BUT Q7. My experience as a tech in situations like this has taught me to suspect capacitors first, transistors next. I've seen transistors fail in some strange ways over the years. :) I've recently learned that disc ceramic caps are not ideal for audio coupling. You guys have got me thinking about distortion and such. I'm thinking about putting the TR-7 on the bench and trying out that nifty surplus distortion analyzer that's languising out in the garage. 73 -Jim The use of ceramic caps in audio circuits is somewhat controversial. Ceramics of some types, mainly high-K types, do tend to have a variation of capacitance with voltage and some audio folks think this contributes some distortion. However, in most coupling circuits there is a steady DC voltage on the cap that exceeds the AC voltage. Also, I have never seen any rigorous research on this showing that a real difference exists. High-K ceramics do have high temperature coeficient of capacitance but low-K ones can have zero temperature variation (NPO types). There can also be a small variation in effective capacitance due to interfacial polarization (don't ask) but mica caps have the same problem. This is of concern only were a cap is used for precision measurement at low frequencies. Paper and plastic film caps have virtually none of this effect. I doubt if it has any significance for normal audio use. Ceramic caps have the virtues of small sizes, availability in high voltages, and great reliability and long life. -- Richard Knoppow Los Angeles WB6KBL dickb...@ix.netcom.com ___ Drakelist mailing list Drakelist@zerobeat.net http://mailman.zerobeat.net/mailman/listinfo/drakelist
Re: [Drakelist] R-4B audio problem traced to low beta Q7 (2N3394)
On Sun, 17 Apr 2011 13:07:35 -0700, Dennis Monticelli wrote: >That would be an interesting study, going stage by stage through the audio >channel of a 4 or 7 series radio. I'll bet that when the R-4B was designed, >the Drake lab didn't have such a piece of equipment and probably went by ear >or visuals on a scope. I've got a couple of HP and a Leader. The Leader gets the nod at the moment because it has a much smaller footprint. I serously need to declutter my workbench... 73 -Jim -- Ham Radio NU0C Lincoln, Nebraska, U.S.S.A. TR7/RV7/R7A/L7, TR6/RV6, T4XC/R4C/L4B, NCL2000, SB104A, R390A, GT550A/RV550A, HyGain 3750, IBM PS/2 - all vintage, all the time! "Give a man a URL, and he will learn for an hour; teach him to Google, and he will learn for a lifetime." HyGain 3750 User's Group - http://groups.yahoo.com/group/HyGain_3750/ http://incolor.inetnebr.com/jshorney http://www.nebraskaghosts.org ___ Drakelist mailing list Drakelist@zerobeat.net http://mailman.zerobeat.net/mailman/listinfo/drakelist
Re: [Drakelist] R-4B audio problem traced to low beta Q7 (2N3394)
Dennis, Late that night when Garey & I decided to swap the transistor, I gave a list of the few I had and he said the 2N3393 would be fine...and he was right on. Probably any General purpose NPN switching/amp transistor would work OK...I don't think it's all that critical of a circuit. 73, Lee -Original Message- From: Dennis Monticelli To: kc9cdt Cc: k4oah ; drakelist Sent: Sun, Apr 17, 2011 4:04 pm Subject: Re: R-4B audio problem traced to low beta Q7 (2N3394) I believe it. However, I think it had less to do with the transistor being a 2N3393 and pretty much all due to just having a beta high enough to allow the transistor stage to bias correctly. Another transistor type with good beta would have probably worked as well. You kept it "in the family" by choosing the 2N3393, so I prefer the route you took. Dennis AE6C On Sun, Apr 17, 2011 at 12:33 PM, wrote: One more thing to addthe 2N3393 worked super (batter than the 2N3394??)...in fact, at a AF gain setting of 8-9 o'clock...the audio is very nice at at a good listening levelat 10:00 with the RF gain down on a noisey band the R-4b really hears well. 73, Lee -Original Message- From: Garey Barrell To: Dennis Monticelli Cc: kc9cdt ; drakelist Sent: Sun, Apr 17, 2011 3:25 pm Subject: Re: R-4B audio problem traced to low beta Q7 (2N3394) Dennis - I think the saying is 'even a blind squirrel finds an acorn now and then, if you put him under an oak tree!" :-) That's why I suggested Beta, a low B-E current just made it worse. I've never seen a small signal BJT fail that way, but your explanation makes sense. I have seen some of the early RF Power transistors that were actually multiple transistors all collected in parallel and one or more would blow open! I believe R141 is ALSO a form of temperature compensation, as the transistor 'warms' up, the B-E drop lowers, increasing the bias current. The drop across the E resistor also increases, decreasing the bias current. Much more stable DC operating point. It DOES also add some negative feedback for the signal, reducing distortion. It also stabilizes the circuit with different transistors as without it the stage gain is limited ONLY by device Beta and internal resistance. With the E resistor, the output is the ratio of Er to Cr, swamping device variations, for a gain of about 580. Very interesting exercise. Thanks very much for your help and expertise! 73, Garey - K4OAH Glen Allen, VA Dennis Monticelli wrote: Lee, Garey, OK. Mystery solved. Curve tracers are wonderful forensic tools. Lee, when I first looked at the 2N3394 from your R-4B it's I-V curves looked fine except for a decided dropoff in low current beta. At first I didn't flag this as a serious issue. The beta of all transistors is a function of collector current, tending to peak right where the factory DC tests are run :-) The falloff rate at low collector current was greater for transistors built over 40 years ago vs a transistor made today. The 2N3394 in your radio still meets min spec of 55 (barely) at 2mA. However, Drake ran the collector current at 1/8 that value and used a base current biasing method (in common use at the time) that suffers from excessive dependency upon the beta value, not for voltage gain but for setting up the correct bias levels. Your particular 2N3394 has a beta of only 25 at 250uA which is why it set up too low. Once it set up too low the further declining beta below 250uA made things even worse, effectively negating the compensating extra current through the 2.2M base bias resistor (the collector voltage rises under the starved collector current condition and would normally help provide addtional base bias). When you placed a fresh 2N3393 in there (min beta of 90 at 2mA), the bias problem went away. Not only was its beta higher at 2mA, it probably fell off a lot less at 250uA also. So the only mystery left is why Q7 worked well enough to leave the factory but not at the present time. Low current beta is dependent upon "surface states" in the Silicon crystal in the region of the base structure. Early transistor manufacturing techniques left much to be desired in terms of surface state control and early plastic packaging (think leaching of ions) compounded the stituation. So I'm guessing the low current beta was marginal when it left the factory and slid down over time. Another posibility is a reverse application of BE voltage (even momentary) as that is known to degrade low current beta. Looking at the circuit the only way that's going to happen is an errant application of a test probe. So most likely it was slow degradation over time. Anyway, the data supports your 2N3393 fix. Given what I've learned I would recommned that concerned R-4B users make a simple DC measurement of the voltage across R141 and if excessively low (i.e. well below 20mV), then do exactly what you did: replace Q7 with a fresh 2N3393 and recheck bias leve
Re: [Drakelist] R-4B audio problem traced to low beta Q7 (2N3394)
That would be an interesting study, going stage by stage through the audio channel of a 4 or 7 series radio. I'll bet that when the R-4B was designed, the Drake lab didn't have such a piece of equipment and probably went by ear or visuals on a scope. Maybe someone familar with what Drake had to work with back then could shed some light on this. Dennis AE6C On Sun, Apr 17, 2011 at 12:59 PM, Jim Shorney wrote: > On Sun, 17 Apr 2011 15:13:23 -0400 (EDT), kc9...@aol.com wrote: > > >Very good explanation...I understoof about 30% of it...but then I am a > >tech., not an engineer. Hi, Hi. > > >What made this tough is that Q7 was kind of working...so I was > >checking, replacing everything BUT Q7. > > My experience as a tech in situations like this has taught me to suspect > capacitors first, transistors next. I've seen transistors fail in some > strange > ways over the years. :) > > I've recently learned that disc ceramic caps are not ideal for audio > coupling. > You guys have got me thinking about distortion and such. I'm thinking about > putting the TR-7 on the bench and trying out that nifty surplus distortion > analyzer that's languising out in the garage. > > 73 > > -Jim > > > -- > Ham Radio NU0C > Lincoln, Nebraska, U.S.S.A. > TR7/RV7/R7A/L7, TR6/RV6, T4XC/R4C/L4B, NCL2000, SB104A, R390A, > GT550A/RV550A, HyGain 3750, IBM PS/2 - all vintage, all the time! > > "Give a man a URL, and he will learn for an hour; teach him to Google, and > he will learn for a lifetime." > > HyGain 3750 User's Group - http://groups.yahoo.com/group/HyGain_3750/ > http://incolor.inetnebr.com/jshorney > http://www.nebraskaghosts.org > > > > ___ > Drakelist mailing list > Drakelist@zerobeat.net > http://mailman.zerobeat.net/mailman/listinfo/drakelist > ___ Drakelist mailing list Drakelist@zerobeat.net http://mailman.zerobeat.net/mailman/listinfo/drakelist
Re: [Drakelist] R-4B audio problem traced to low beta Q7 (2N3394)
I believe it. However, I think it had less to do with the transistor being a 2N3393 and pretty much all due to just having a beta high enough to allow the transistor stage to bias correctly. Another transistor type with good beta would have probably worked as well. You kept it "in the family" by choosing the 2N3393, so I prefer the route you took. Dennis AE6C On Sun, Apr 17, 2011 at 12:33 PM, wrote: > One more thing to addthe 2N3393 worked super (batter than the > 2N3394??)...in fact, at a AF gain setting of 8-9 o'clock...the audio is very > nice at at a good listening levelat 10:00 with the RF gain down on a > noisey band the R-4b really hears well. > 73, > Lee > > > > > -Original Message- > From: Garey Barrell > To: Dennis Monticelli > Cc: kc9cdt ; drakelist > Sent: Sun, Apr 17, 2011 3:25 pm > Subject: Re: R-4B audio problem traced to low beta Q7 (2N3394) > > > Dennis - > > I think the saying is 'even a blind squirrel finds an acorn now and > then, if you put him under an oak tree!" :-) > > That's why I suggested Beta, a low B-E current just made it worse. I've > never seen a small signal BJT fail that way, but your explanation makes > sense. I have seen some of the early RF Power transistors that were > actually multiple transistors all collected in parallel and one or more > would blow open! > > I believe R141 is ALSO a form of temperature compensation, as the > transistor 'warms' up, the B-E drop lowers, increasing the bias > current. The drop across the E resistor also increases, decreasing the > bias current. Much more stable DC operating point. It DOES also add > some negative feedback for the signal, reducing distortion. It also > stabilizes the circuit with different transistors as without it the > stage gain is limited ONLY by device Beta and internal resistance. With > the E resistor, the output is the ratio of Er to Cr, swamping device > variations, for a gain of about 580. > > Very interesting exercise. Thanks very much for your help and expertise! > > 73, Garey - K4OAH > Glen Allen, VA > > > > Dennis Monticelli wrote: > >> Lee, Garey, >> OK. Mystery solved. Curve tracers are wonderful forensic tools. >> Lee, when I first looked at the 2N3394 from your R-4B it's I-V curves >> looked fine except for a decided dropoff in low current beta. At >> first I didn't flag this as a serious issue. The beta of all >> transistors is a function of collector current, tending to peak right >> where the factory DC tests are run :-) The falloff rate at low >> collector current was greater for transistors built over 40 years ago >> vs a transistor made today. The 2N3394 in your radio still meets min >> spec of 55 (barely) at 2mA. However, Drake ran the collector current >> at 1/8 that value and used a base current biasing method (in common >> use at the time) that suffers from excessive dependency upon the beta >> value, not for voltage gain but for setting up the correct bias >> levels. Your particular 2N3394 has a beta of only 25 at 250uA which >> is why it set up too low. Once it set up too low the further >> declining beta below 250uA made things even worse, effectively >> negating the compensating extra current through the 2.2M base bias >> resistor (the collector voltage rises under the starved collector >> current condition and would normally help provide addtional base >> bias). When you placed a fresh 2N3393 in there (min beta of 90 at >> 2mA), the bias problem went away. Not only was its beta higher at >> 2mA, it probably fell off a lot less at 250uA also. >> So the only mystery left is why Q7 worked well enough to leave the >> factory but not at the present time. Low current beta is dependent >> upon "surface states" in the Silicon crystal in the region of the >> > base > >> structure. Early transistor manufacturing techniques left much to be >> desired in terms of surface state control and early plastic packaging >> (think leaching of ions) compounded the stituation. So I'm guessing >> the low current beta was marginal when it left the factory and slid >> down over time. Another posibility is a reverse application of BE >> voltage (even momentary) as that is known to degrade low current >> beta. Looking at the circuit the only way that's going to happen is >> an errant application of a test probe. So most likely it was slow >> degradation over time. >> Anyway, the data supports your 2N3393 fix. Given what I've learned I >> would recommned that concerned R-4B users make a simple DC >> > measurement > >> of the voltage across R141 and if excessively low (i.e. well below >> 20mV), then do exactly what you did: replace Q7 with a fresh 2N3393 >> and recheck bias levels. R-4B owners may want to do this because >> > R141 > >> is not there to manage bias, it is there for emitter degeneration, a >> form of negative feedback that reduces distortion under strong audio >> signals. If insufficient voltage is developed across R141 the >> distortion will be great
Re: [Drakelist] R-4B audio problem traced to low beta Q7 (2N3394)
Lee, Your experience of the having audio that was partially working with evidence of low gain and distortion jives with a starved bias situation. Thanks for diving into this problem, Lee, the group has learned something useful from it. Garey, Yes, I remember you suspecting low beta. It turned out to be right! As for the emitter resistor, R141, it really doesn't impact the biasing either at room or another temp. Whether R141's drop is small (20mV) or 10X larger because 820 ohm was used instead, the bias feed current through the 2.2M is virtually the same. That's because the base biasing scheme is current fed from a base resistor that has many volts across it. Now had Drake applied VOLTAGE bias on the base, then it would have made a difference. Regardless of bias the emitter resistor does contribute to greater uniformity of input resistance and voltage gain from device to device and over temp as you say. The amount of the voltage drop also contributes to greater linearity of gain (i.e lower distortion). To roughly estimate the degree of linearity improvement simply divide the emitter resistor voltage drop by KT/q, which is roughly 25mV. Thus the normal drop across R141 (20mV) improves linearity by approximately a factor of two. Doubling R141 would help linearity further but simultaneously reduces the votage gain of the stage by about the same factor. I would not be surprised to learn that the Drake circuit designer started out with no emitter resistance and then noticed his distorition was a little higher than he would prefer, yet he had some gain to spare.thus R141 with only a tad of voltage across it was born. As for unit cells in power transistors blowing open, what happens is first is a melt down of the Silicon within the unit cell causing a short and then the fusing open of emitter or collector metalization feeding that cell. Depending upon how that metalization is routed a portion of that transistor may blow open or the entire current path may open should a main metal branch blow. SOA (Safe Operating Area) failure is rare in small signal transistors and common in power transistors. While the SOA mechanism in present in both types of devices, circuits tend to stress power transistors to the limits, thus exposing that failure mode. Dennis AE6C On Sun, Apr 17, 2011 at 12:25 PM, Garey Barrell wrote: > Dennis - > > I think the saying is 'even a blind squirrel finds an acorn now and then, > if you put him under an oak tree!" :-) > > That's why I suggested Beta, a low B-E current just made it worse. I've > never seen a small signal BJT fail that way, but your explanation makes > sense. I have seen some of the early RF Power transistors that were > actually multiple transistors all collected in parallel and one or more > would blow open! > > I believe R141 is ALSO a form of temperature compensation, as the > transistor 'warms' up, the B-E drop lowers, increasing the bias current. > The drop across the E resistor also increases, decreasing the bias current. > Much more stable DC operating point. It DOES also add some negative > feedback for the signal, reducing distortion. It also stabilizes the > circuit with different transistors as without it the stage gain is limited > ONLY by device Beta and internal resistance. With the E resistor, the > output is the ratio of Er to Cr, swamping device variations, for a gain of > about 580. > > Very interesting exercise. Thanks very much for your help and expertise! > > 73, Garey - K4OAH > Glen Allen, VA > > > > > Dennis Monticelli wrote: > >> Lee, Garey, >> OK. Mystery solved. Curve tracers are wonderful forensic tools. >> Lee, when I first looked at the 2N3394 from your R-4B it's I-V curves >> looked fine except for a decided dropoff in low current beta. At first I >> didn't flag this as a serious issue. The beta of all transistors is a >> function of collector current, tending to peak right where the factory DC >> tests are run :-) The falloff rate at low collector current was greater for >> transistors built over 40 years ago vs a transistor made today. The 2N3394 >> in your radio still meets min spec of 55 (barely) at 2mA. However, Drake >> ran the collector current at 1/8 that value and used a base current biasing >> method (in common use at the time) that suffers from excessive dependency >> upon the beta value, not for voltage gain but for setting up the correct >> bias levels. Your particular 2N3394 has a beta of only 25 at 250uA which is >> why it set up too low. Once it set up too low the further declining beta >> below 250uA made things even worse, effectively negating the compensating >> extra current through the 2.2M base bias resistor (the collector voltage >> rises under the starved collector current condition and would normally help >> provide addtional base bias). When you placed a fresh 2N3393 in there (min >> beta of 90 at 2mA), the bias problem went away. Not only was its beta >> higher at 2mA, it probably fell
Re: [Drakelist] R-4B audio problem traced to low beta Q7 (2N3394)
On Sun, 17 Apr 2011 15:13:23 -0400 (EDT), kc9...@aol.com wrote: >Very good explanation...I understoof about 30% of it...but then I am a >tech., not an engineer. Hi, Hi. >What made this tough is that Q7 was kind of working...so I was >checking, replacing everything BUT Q7. My experience as a tech in situations like this has taught me to suspect capacitors first, transistors next. I've seen transistors fail in some strange ways over the years. :) I've recently learned that disc ceramic caps are not ideal for audio coupling. You guys have got me thinking about distortion and such. I'm thinking about putting the TR-7 on the bench and trying out that nifty surplus distortion analyzer that's languising out in the garage. 73 -Jim -- Ham Radio NU0C Lincoln, Nebraska, U.S.S.A. TR7/RV7/R7A/L7, TR6/RV6, T4XC/R4C/L4B, NCL2000, SB104A, R390A, GT550A/RV550A, HyGain 3750, IBM PS/2 - all vintage, all the time! "Give a man a URL, and he will learn for an hour; teach him to Google, and he will learn for a lifetime." HyGain 3750 User's Group - http://groups.yahoo.com/group/HyGain_3750/ http://incolor.inetnebr.com/jshorney http://www.nebraskaghosts.org ___ Drakelist mailing list Drakelist@zerobeat.net http://mailman.zerobeat.net/mailman/listinfo/drakelist
Re: [Drakelist] R-4B audio problem traced to low beta Q7 (2N3394)
One more thing to addthe 2N3393 worked super (batter than the 2N3394??)...in fact, at a AF gain setting of 8-9 o'clock...the audio is very nice at at a good listening levelat 10:00 with the RF gain down on a noisey band the R-4b really hears well. 73, Lee -Original Message- From: Garey Barrell To: Dennis Monticelli Cc: kc9cdt ; drakelist Sent: Sun, Apr 17, 2011 3:25 pm Subject: Re: R-4B audio problem traced to low beta Q7 (2N3394) Dennis - I think the saying is 'even a blind squirrel finds an acorn now and then, if you put him under an oak tree!" :-) That's why I suggested Beta, a low B-E current just made it worse. I've never seen a small signal BJT fail that way, but your explanation makes sense. I have seen some of the early RF Power transistors that were actually multiple transistors all collected in parallel and one or more would blow open! I believe R141 is ALSO a form of temperature compensation, as the transistor 'warms' up, the B-E drop lowers, increasing the bias current. The drop across the E resistor also increases, decreasing the bias current. Much more stable DC operating point. It DOES also add some negative feedback for the signal, reducing distortion. It also stabilizes the circuit with different transistors as without it the stage gain is limited ONLY by device Beta and internal resistance. With the E resistor, the output is the ratio of Er to Cr, swamping device variations, for a gain of about 580. Very interesting exercise. Thanks very much for your help and expertise! 73, Garey - K4OAH Glen Allen, VA Dennis Monticelli wrote: Lee, Garey, OK. Mystery solved. Curve tracers are wonderful forensic tools. Lee, when I first looked at the 2N3394 from your R-4B it's I-V curves looked fine except for a decided dropoff in low current beta. At first I didn't flag this as a serious issue. The beta of all transistors is a function of collector current, tending to peak right where the factory DC tests are run :-) The falloff rate at low collector current was greater for transistors built over 40 years ago vs a transistor made today. The 2N3394 in your radio still meets min spec of 55 (barely) at 2mA. However, Drake ran the collector current at 1/8 that value and used a base current biasing method (in common use at the time) that suffers from excessive dependency upon the beta value, not for voltage gain but for setting up the correct bias levels. Your particular 2N3394 has a beta of only 25 at 250uA which is why it set up too low. Once it set up too low the further declining beta below 250uA made things even worse, effectively negating the compensating extra current through the 2.2M base bias resistor (the collector voltage rises under the starved collector current condition and would normally help provide addtional base bias). When you placed a fresh 2N3393 in there (min beta of 90 at 2mA), the bias problem went away. Not only was its beta higher at 2mA, it probably fell off a lot less at 250uA also. So the only mystery left is why Q7 worked well enough to leave the factory but not at the present time. Low current beta is dependent upon "surface states" in the Silicon crystal in the region of the base structure. Early transistor manufacturing techniques left much to be desired in terms of surface state control and early plastic packaging (think leaching of ions) compounded the stituation. So I'm guessing the low current beta was marginal when it left the factory and slid down over time. Another posibility is a reverse application of BE voltage (even momentary) as that is known to degrade low current beta. Looking at the circuit the only way that's going to happen is an errant application of a test probe. So most likely it was slow degradation over time. Anyway, the data supports your 2N3393 fix. Given what I've learned I would recommned that concerned R-4B users make a simple DC measurement of the voltage across R141 and if excessively low (i.e. well below 20mV), then do exactly what you did: replace Q7 with a fresh 2N3393 and recheck bias levels. R-4B owners may want to do this because R141 is not there to manage bias, it is there for emitter degeneration, a form of negative feedback that reduces distortion under strong audio signals. If insufficient voltage is developed across R141 the distortion will be greater. Hope this explanation helps the list. I learned something here. Dennis AE6C ___ Drakelist mailing list Drakelist@zerobeat.net http://mailman.zerobeat.net/mailman/listinfo/drakelist
Re: [Drakelist] R-4B audio problem traced to low beta Q7 (2N3394)
Dennis - I think the saying is 'even a blind squirrel finds an acorn now and then, if you put him under an oak tree!" :-) That's why I suggested Beta, a low B-E current just made it worse. I've never seen a small signal BJT fail that way, but your explanation makes sense. I have seen some of the early RF Power transistors that were actually multiple transistors all collected in parallel and one or more would blow open! I believe R141 is ALSO a form of temperature compensation, as the transistor 'warms' up, the B-E drop lowers, increasing the bias current. The drop across the E resistor also increases, decreasing the bias current. Much more stable DC operating point. It DOES also add some negative feedback for the signal, reducing distortion. It also stabilizes the circuit with different transistors as without it the stage gain is limited ONLY by device Beta and internal resistance. With the E resistor, the output is the ratio of Er to Cr, swamping device variations, for a gain of about 580. Very interesting exercise. Thanks very much for your help and expertise! 73, Garey - K4OAH Glen Allen, VA Dennis Monticelli wrote: Lee, Garey, OK. Mystery solved. Curve tracers are wonderful forensic tools. Lee, when I first looked at the 2N3394 from your R-4B it's I-V curves looked fine except for a decided dropoff in low current beta. At first I didn't flag this as a serious issue. The beta of all transistors is a function of collector current, tending to peak right where the factory DC tests are run :-) The falloff rate at low collector current was greater for transistors built over 40 years ago vs a transistor made today. The 2N3394 in your radio still meets min spec of 55 (barely) at 2mA. However, Drake ran the collector current at 1/8 that value and used a base current biasing method (in common use at the time) that suffers from excessive dependency upon the beta value, not for voltage gain but for setting up the correct bias levels. Your particular 2N3394 has a beta of only 25 at 250uA which is why it set up too low. Once it set up too low the further declining beta below 250uA made things even worse, effectively negating the compensating extra current through the 2.2M base bias resistor (the collector voltage rises under the starved collector current condition and would normally help provide addtional base bias). When you placed a fresh 2N3393 in there (min beta of 90 at 2mA), the bias problem went away. Not only was its beta higher at 2mA, it probably fell off a lot less at 250uA also. So the only mystery left is why Q7 worked well enough to leave the factory but not at the present time. Low current beta is dependent upon "surface states" in the Silicon crystal in the region of the base structure. Early transistor manufacturing techniques left much to be desired in terms of surface state control and early plastic packaging (think leaching of ions) compounded the stituation. So I'm guessing the low current beta was marginal when it left the factory and slid down over time. Another posibility is a reverse application of BE voltage (even momentary) as that is known to degrade low current beta. Looking at the circuit the only way that's going to happen is an errant application of a test probe. So most likely it was slow degradation over time. Anyway, the data supports your 2N3393 fix. Given what I've learned I would recommned that concerned R-4B users make a simple DC measurement of the voltage across R141 and if excessively low (i.e. well below 20mV), then do exactly what you did: replace Q7 with a fresh 2N3393 and recheck bias levels. R-4B owners may want to do this because R141 is not there to manage bias, it is there for emitter degeneration, a form of negative feedback that reduces distortion under strong audio signals. If insufficient voltage is developed across R141 the distortion will be greater. Hope this explanation helps the list. I learned something here. Dennis AE6C ___ Drakelist mailing list Drakelist@zerobeat.net http://mailman.zerobeat.net/mailman/listinfo/drakelist
Re: [Drakelist] R-4B audio problem traced to low beta Q7 (2N3394)
Dennis, Very good explanation...I understoof about 30% of it...but then I am a tech., not an engineer. Hi, Hi. Since my friend had loaned the R-4B out for a few years, he does not know when the problem started, or if it was gradual or all of a sudden. When we were working on it...Garey rightly so kept telling me the voltages are wrong on Q7...it proved to be correct. What made this tough is that Q7 was kind of working...so I was checking, replacing everything BUT Q7. When Q7 was correctly replaced the R-4B sprang to life. (Another lesson...don't change parts out at 1:00AM! Hi, Hi) One thing for everyone here to understand is: The audio was still wotking, just not wellsome may have just thought well, we have a weak tube or something and used it the way it was! The thing is at the 12:00 position of the AF gain...it was useable...just not right. Noe at 12:00 it will blow you away in volume...and when it was not correct there was some distortion appearing especially after 12:00 on the gain. Thanks again go out to Garey for hanging in there with me to get this fixed, and to you Dennis for letting us know what was actually wrong with Q7. So, to everyone in Drakeland, watch your P's & Q7's 73, Lee, KC9CDT -Original Message- From: Dennis Monticelli To: kc9cdt ; Garey Barrell Cc: drakelist Sent: Sun, Apr 17, 2011 2:52 pm Subject: R-4B audio problem traced to low beta Q7 (2N3394) Lee, Garey, OK. Mystery solved. Curve tracers are wonderful forensic tools. Lee, when I first looked at the 2N3394 from your R-4B it's I-V curves looked fine except for a decided dropoff in low current beta. At first I didn't flag this as a serious issue. The beta of all transistors is a function of collector current, tending to peak right where the factory DC tests are run :-) The falloff rate at low collector current was greater for transistors built over 40 years ago vs a transistor made today. The 2N3394 in your radio still meets min spec of 55 (barely) at 2mA. However, Drake ran the collector current at 1/8 that value and used a base current biasing method (in common use at the time) that suffers from excessive dependency upon the beta value, not for voltage gain but for setting up the correct bias levels. Your particular 2N3394 has a beta of only 25 at 250uA which is why it set up too low. Once it set up too low the further declining beta below 250uA made things even worse, effectively negating the compensating extra current through the 2.2M base bias resistor (the collector voltage rises under the starved collector current condition and would normally help provide addtional base bias). When you placed a fresh 2N3393 in there (min beta of 90 at 2mA), the bias problem went away. Not only was its beta higher at 2mA, it probably fell off a lot less at 250uA also. So the only mystery left is why Q7 worked well enough to leave the factory but not at the present time. Low current beta is dependent upon "surface states" in the Silicon crystal in the region of the base structure. Early transistor manufacturing techniques left much to be desired in terms of surface state control and early plastic packaging (think leaching of ions) compounded the stituation. So I'm guessing the low current beta was marginal when it left the factory and slid down over time. Another posibility is a reverse application of BE voltage (even momentary) as that is known to degrade low current beta. Looking at the circuit the only way that's going to happen is an errant application of a test probe. So most likely it was slow degradation over time. Anyway, the data supports your 2N3393 fix. Given what I've learned I would recommned that concerned R-4B users make a simple DC measurement of the voltage across R141 and if excessively low (i.e. well below 20mV), then do exactly what you did: replace Q7 with a fresh 2N3393 and recheck bias levels. R-4B owners may want to do this because R141 is not there to manage bias, it is there for emitter degeneration, a form of negative feedback that reduces distortion under strong audio signals. If insufficient voltage is developed across R141 the distortion will be greater. Hope this explanation helps the list. I learned something here. Dennis AE6C ___ Drakelist mailing list Drakelist@zerobeat.net http://mailman.zerobeat.net/mailman/listinfo/drakelist
[Drakelist] R-4B audio problem traced to low beta Q7 (2N3394)
Lee, Garey, OK. Mystery solved. Curve tracers are wonderful forensic tools. Lee, when I first looked at the 2N3394 from your R-4B it's I-V curves looked fine except for a decided dropoff in low current beta. At first I didn't flag this as a serious issue. The beta of all transistors is a function of collector current, tending to peak right where the factory DC tests are run :-) The falloff rate at low collector current was greater for transistors built over 40 years ago vs a transistor made today. The 2N3394 in your radio still meets min spec of 55 (barely) at 2mA. However, Drake ran the collector current at 1/8 that value and used a base current biasing method (in common use at the time) that suffers from excessive dependency upon the beta value, not for voltage gain but for setting up the correct bias levels. Your particular 2N3394 has a beta of only 25 at 250uA which is why it set up too low. Once it set up too low the further declining beta below 250uA made things even worse, effectively negating the compensating extra current through the 2.2M base bias resistor (the collector voltage rises under the starved collector current condition and would normally help provide addtional base bias). When you placed a fresh 2N3393 in there (min beta of 90 at 2mA), the bias problem went away. Not only was its beta higher at 2mA, it probably fell off a lot less at 250uA also. So the only mystery left is why Q7 worked well enough to leave the factory but not at the present time. Low current beta is dependent upon "surface states" in the Silicon crystal in the region of the base structure. Early transistor manufacturing techniques left much to be desired in terms of surface state control and early plastic packaging (think leaching of ions) compounded the stituation. So I'm guessing the low current beta was marginal when it left the factory and slid down over time. Another posibility is a reverse application of BE voltage (even momentary) as that is known to degrade low current beta. Looking at the circuit the only way that's going to happen is an errant application of a test probe. So most likely it was slow degradation over time. Anyway, the data supports your 2N3393 fix. Given what I've learned I would recommned that concerned R-4B users make a simple DC measurement of the voltage across R141 and if excessively low (i.e. well below 20mV), then do exactly what you did: replace Q7 with a fresh 2N3393 and recheck bias levels. R-4B owners may want to do this because R141 is not there to manage bias, it is there for emitter degeneration, a form of negative feedback that reduces distortion under strong audio signals. If insufficient voltage is developed across R141 the distortion will be greater. Hope this explanation helps the list. I learned something here. Dennis AE6C ___ Drakelist mailing list Drakelist@zerobeat.net http://mailman.zerobeat.net/mailman/listinfo/drakelist