Re: Analog to Digital Converter
On 8/18/2015 7:01 PM, dwight wrote: Reading the note made me think of something else. If it is an SAR, and you read to soon, the LSBs of the incomplete conversion will always be the same. All ones or all zeros. If it is a Sigma-Delta, and the input has noise on it that is above the same rate, the LSB's will take on a random, all ones or all zeros on different reading. This is an indication that the input has not been properly filtered or you have exceeded the slew rate, even if you are within the frequency band width. Most audio A/Ds assume a de-emphesis so that the higher frequencies are reduced in slew rate. Any noise above 1/2 the sample rate will be reflected into the values. Anything above the sample rate will be as I described, a mix of all 1's or all 0's on each sample. Dwight That is exactly what I am seeing. The signal the ADC is trying to capture looks like a sinc function, lots of small ripples with a narrow high amplitude burst in the center. What gets lost are the small amplitude values. Doug
Re: Analog to Digital Converter
On 8/15/2015 12:40 PM, tony duell wrote: I have a number of laboratory instruments that are from the 1990 time frame. They produce digital data that is the digitized signal from a detector, the data can be from 512 to 65K samples long. The ADC used in these instruments is a 16bit 100ksample/sec design. The ADC is in a 3 by 4 inch metal box with a row of pins on each long edge. [...] What is inside the box? Is it a hybrid circuit? I came across somewhat similar looking ADC and DAC modules in an I2S image display system. These were flat metal cans with pins on the bottom, going into individual sockets on the PCB. In that case the can could be opened up quite easily (I think just a couple of points to unsolder. Inside was a PCB _stuffed_ with components, including several possibly custom metal-can ICs (in the case of the ADC I susect fast analogue comparators). Have you tried to open your module? It may not be potted. -tony I got a close enough look to get some markings deciphered. It is an Analog Solutions ZAD 2716-1 model. A picture of it can be found here: http://www.stronic.fr/photos/ANA1001.jpg The metal box is a shield and is soldered to the circuit card, I think that opening it up would be sort of a one way trip. I found a discussion of it at: http://www.lavryengineering.com/lavry_forum/viewtopic.php?t=598 The fellow who designed it actually has some kind of audio forum! Doug
Re: Analog to Digital Converter
On 08/15/2015 11:18 AM, Douglas Taylor wrote: I have a number of laboratory instruments that are from the 1990 time frame. They produce digital data that is the digitized signal from a detector, the data can be from 512 to 65K samples long. The ADC used in these instruments is a 16bit 100ksample/sec design. The ADC is in a 3 by 4 inch metal box with a row of pins on each long edge. I think some of them are failing because I get the full 16 bit resolution from one machine, but not the others. This was determined by taking the digital samples and sorting the values and computing the increments between the adjacent values. In some cases the output looked like 14 bit resolution and in one case 6 bit resolution. Does anyone have any experience with technology? Who was the manufacturer? (There is no id on the outside) What is inside the box? Is it a hybrid circuit? DDC made a number of hybrid ADCs, but I've never seen one that was 3 x 4! That's really big. Most likely there is a normal PC board inside, with a bunch of chips and discrete components on it. Is this box hermetic, or maybe if you peel back a label, there may be screws that hold it together. Jon
RE: Analog to Digital Converter
Most of these older module use successive approximation converters. If you read them too fast, you'll only get a partial conversion. A number of manufactures made these modules. Analog Devices, Harris, Beckman and several others. Dwight From: a...@p850ug1.demon.co.uk To: cct...@classiccmp.org Subject: RE: Analog to Digital Converter Date: Sat, 15 Aug 2015 16:40:40 + I have a number of laboratory instruments that are from the 1990 time frame. They produce digital data that is the digitized signal from a detector, the data can be from 512 to 65K samples long. The ADC used in these instruments is a 16bit 100ksample/sec design. The ADC is in a 3 by 4 inch metal box with a row of pins on each long edge. [...] What is inside the box? Is it a hybrid circuit? I came across somewhat similar looking ADC and DAC modules in an I2S image display system. These were flat metal cans with pins on the bottom, going into individual sockets on the PCB. In that case the can could be opened up quite easily (I think just a couple of points to unsolder. Inside was a PCB _stuffed_ with components, including several possibly custom metal-can ICs (in the case of the ADC I susect fast analogue comparators). Have you tried to open your module? It may not be potted. -tony
RE: Analog to Digital Converter
I have a number of laboratory instruments that are from the 1990 time frame. They produce digital data that is the digitized signal from a detector, the data can be from 512 to 65K samples long. The ADC used in these instruments is a 16bit 100ksample/sec design. The ADC is in a 3 by 4 inch metal box with a row of pins on each long edge. [...] What is inside the box? Is it a hybrid circuit? I came across somewhat similar looking ADC and DAC modules in an I2S image display system. These were flat metal cans with pins on the bottom, going into individual sockets on the PCB. In that case the can could be opened up quite easily (I think just a couple of points to unsolder. Inside was a PCB _stuffed_ with components, including several possibly custom metal-can ICs (in the case of the ADC I susect fast analogue comparators). Have you tried to open your module? It may not be potted. -tony
RE: Analog to Digital Converter
At 12:56 AM 8/16/2015, Dwight wrote: Most of these older module use successive approximation converters. If you read them too fast, you'll only get a partial conversion. I'll second that. The ones I have seen most often (in another industry) were Harris. For those who would like who would like to increase their understanding of DSP I suggest: http://what-when-how.com/Tutorial/topic-3062l7iib/Applications-of-DSP-to-Audio-and-Acoustics-113.html For greater background on and history of DSP I suggest looking up some of the references cited in that paper if you have access to a good EE library. Dale H. Cook, Roanoke/Lynchburg, VA Osborne 1 / Kaypro 4-84 / Kaypro 1 / Amstrad PPC-640 http://plymouthcolony.net/starcity/radios/index.html
RE: Analog to Digital Converter
I was thinking that around this time, dual slop detectors were becoming popular but it is unlikely this is your problem. If reading a dual slope before conversion is complete, the value read would be low in value, not just missing LSBs. They were generally slower as well but often in the 16 or more bits of accuracy. Tinker Dwight
Re: Analog to Digital Converter
On 8/15/15 6:41 PM, Jon Elson wrote: DDC made a number of hybrid ADCs, but I've never seen one that was 3 x 4! That's really big. Some of the Data Translation modules were that big. The normally had the block diagram / part number / and Data Translation silk screened on the top of them. ADAC made some big modules as well. DT boards were pretty common in Multibus and DEC versions.
Re: Analog to Digital Converter
I believe that is true, it is a SAR type of ADC. That is an import piece of information there concerning reading the ADC too fast. Because I have one instrument that is working properly I can use that as a baseline to look at how fast the ADC is being pinged on. This gives me something to go on. On 8/16/2015 12:56 AM, dwight wrote: Most of these older module use successive approximation converters. If you read them too fast, you'll only get a partial conversion. A number of manufactures made these modules. Analog Devices, Harris, Beckman and several others. Dwight From: a...@p850ug1.demon.co.uk To: cct...@classiccmp.org Subject: RE: Analog to Digital Converter Date: Sat, 15 Aug 2015 16:40:40 + I have a number of laboratory instruments that are from the 1990 time frame. They produce digital data that is the digitized signal from a detector, the data can be from 512 to 65K samples long. The ADC used in these instruments is a 16bit 100ksample/sec design. The ADC is in a 3 by 4 inch metal box with a row of pins on each long edge. [...] What is inside the box? Is it a hybrid circuit? I came across somewhat similar looking ADC and DAC modules in an I2S image display system. These were flat metal cans with pins on the bottom, going into individual sockets on the PCB. In that case the can could be opened up quite easily (I think just a couple of points to unsolder. Inside was a PCB _stuffed_ with components, including several possibly custom metal-can ICs (in the case of the ADC I susect fast analogue comparators). Have you tried to open your module? It may not be potted. -tony
Re: Analog to Digital Converter
On 8/15/2015 12:40 PM, tony duell wrote: I have a number of laboratory instruments that are from the 1990 time frame. They produce digital data that is the digitized signal from a detector, the data can be from 512 to 65K samples long. The ADC used in these instruments is a 16bit 100ksample/sec design. The ADC is in a 3 by 4 inch metal box with a row of pins on each long edge. [...] What is inside the box? Is it a hybrid circuit? I came across somewhat similar looking ADC and DAC modules in an I2S image display system. These were flat metal cans with pins on the bottom, going into individual sockets on the PCB. In that case the can could be opened up quite easily (I think just a couple of points to unsolder. Inside was a PCB _stuffed_ with components, including several possibly custom metal-can ICs (in the case of the ADC I susect fast analogue comparators). Have you tried to open your module? It may not be potted. -tony The ADC module is difficult to get at, the instrument it is in is very heavy and sealed against moisture. I was hoping to get more of heads up before I go back in.
Re: Analog to Digital Converter
On 8/16/2015 8:24 AM, Al Kossow wrote: On 8/15/15 6:41 PM, Jon Elson wrote: DDC made a number of hybrid ADCs, but I've never seen one that was 3 x 4! That's really big. Some of the Data Translation modules were that big. The normally had the block diagram / part number / and Data Translation silk screened on the top of them. ADAC made some big modules as well. DT boards were pretty common in Multibus and DEC versions. I remember the Data Translation modules, I think the one I'm dealing with is from the same era.