First of all,my thanks to all the contributors to this thread for sharing your knowledge and being such good mentors for noobs like me. Special thanks to Anatech, Cogeniac and Phloodpants for posting their thoughts, schematics, and actual board layouts.
I had 5 boards made using a slightly modified version of of Phloodpants' Eagle files found on post #259 (removed the fan footprint and text Credit text) I have 4 extra boards. PM me, if anyone wants them for the cost of shipping. (I got a discount deal for my first order from JLCPCB) I figure that's the least I can do to give a little back to the community that has taught me so much.
All the photos below show measurements across the collector voltage difference between transistors A and B (2N5401 PNP's) at 1.9 mA measured base current in both A and B transistors (jumper 3 on the PNP DIP swich resistor selector at bottom left of the photos). I like the ability to check the base currents using the jumpers provided on this version of the board! Resistors used at the DIP switches are (L to R) 154r, 316r, 620r, 1.24K, 2.49K, 4.99K, 10K and 20K. I can tell this is a sensitive jig, all I have to do to change readings is touch one of the transistors, and the voltage difference either increases or decreases rapidly, only to settle back once the DUT temps have stabilized. So far so good!
I had 5 boards made using a slightly modified version of of Phloodpants' Eagle files found on post #259 (removed the fan footprint and text Credit text) I have 4 extra boards. PM me, if anyone wants them for the cost of shipping. (I got a discount deal for my first order from JLCPCB) I figure that's the least I can do to give a little back to the community that has taught me so much.
All the photos below show measurements across the collector voltage difference between transistors A and B (2N5401 PNP's) at 1.9 mA measured base current in both A and B transistors (jumper 3 on the PNP DIP swich resistor selector at bottom left of the photos). I like the ability to check the base currents using the jumpers provided on this version of the board! Resistors used at the DIP switches are (L to R) 154r, 316r, 620r, 1.24K, 2.49K, 4.99K, 10K and 20K. I can tell this is a sensitive jig, all I have to do to change readings is touch one of the transistors, and the voltage difference either increases or decreases rapidly, only to settle back once the DUT temps have stabilized. So far so good!
- The first photo (below left) shows the difference in collector voltages between one pair of transistors that was preselected with the same HFE (158, using this meter's HFE function) after sitting for over 5 minutes, while the voltage settled. I switched the transistors from A-to-B and the voltage difference was the same but negative.
- The second photo (below center) shows that adding an insulated copper shroud (a plumbing fitting with foam around the inside and top) after the voltage settled shows a minor difference in voltage from the unshrouded measurement, I believe it mostly ensures the air currents are not having an adverse impact.
- The third photo (below right) shows the voltage difference between one of the initial transistors and another with an HFE of 176. This proves to me that pre-sorting according to HFE saves time and gives me a better starting point for finding matching transistors.
- Is it a problem that I don't have a Zero volt supply? I did move a battery lead from the negative supply to the neutral, and the difference in voltage dropped significantly in all cases, even though when that happens, the right side LED turns off (I would have thought that by powering only one side, the voltage and current would be higher, not lower - but maybe that's what's causing the collector voltages to get closer to each other?)
- I understand that by comparing collector voltages between transistors A and B I'm looking for zero volt differences. Is the difference in voltage I see at a steady room temperature of 20c or 68F (with or without shroud) a close-enough indicator of the true collector voltage match between the two transistors, or is it worth trying to get the transistors to a more realistic (higher) steady operating temperature around 40c or 104F?
- Once I had a pair close enough (.3mV collector voltage difference @ 1.9 mA base current) I tested various jumpers at the resistor selector to see what happened when Base current increased or decreased). When I reduced resistance by one jumper position (#3 to #2) the difference from A to B collector voltages almost doubled. When I increased resistance by (#3 to #4) the the difference from A to B collector voltages almost halved. Is this what others have experienced?
Attachments
Hi Sixto,
Excellent, and thanks for offering the extras to members. That is really good of you!
Preselecting helps a lot. I do that.
1. Yes, you really need a bipolar supply. Two 9V batteries are needed, connect the common to the common on the board. I use a bipolar power supply set around 11 VDC, but bipolar 9V is perfectly fine.
2. Yes it is. You may see some shift as you move on the curves, but no matter how you match them, this will occur. I designed it with selectable tail currents so you could test them at the current they will be running at. You should have the transistors in direct contact with each other and shielded from the air, then shield the entire jig from air currents.
3. Of course, makes sense. You are changing the collector currents, so your measured values will change for sure. The percentage match should remain close.
Excellent, and thanks for offering the extras to members. That is really good of you!
Preselecting helps a lot. I do that.
1. Yes, you really need a bipolar supply. Two 9V batteries are needed, connect the common to the common on the board. I use a bipolar power supply set around 11 VDC, but bipolar 9V is perfectly fine.
2. Yes it is. You may see some shift as you move on the curves, but no matter how you match them, this will occur. I designed it with selectable tail currents so you could test them at the current they will be running at. You should have the transistors in direct contact with each other and shielded from the air, then shield the entire jig from air currents.
3. Of course, makes sense. You are changing the collector currents, so your measured values will change for sure. The percentage match should remain close.
Thank you for your response! i have also learned first-hand what you’ve said about getting the right connectors…
I keep bending transistor leads trying to seat them into the connectors I purchased, (time-consuming and frustrating)
Upon further research, my transistor leads are .56 mm square, and most female connectors I see are designed for inserting.40 to .50 mm leads. (Even ones that cost $3 apiece) I now wish I had modified the board design to use ZIF sockets…
Does anyone have a source/spec for female socket connectors that accept thicker leads???
I can seat the transistors in the current sockets with my hands, by wiggling and using some force, but I’d rather use tweezers to handle them, and so far, that’s not working well…
thanks, Sixto
I keep bending transistor leads trying to seat them into the connectors I purchased, (time-consuming and frustrating)
Upon further research, my transistor leads are .56 mm square, and most female connectors I see are designed for inserting.40 to .50 mm leads. (Even ones that cost $3 apiece) I now wish I had modified the board design to use ZIF sockets…
Does anyone have a source/spec for female socket connectors that accept thicker leads???
I can seat the transistors in the current sockets with my hands, by wiggling and using some force, but I’d rather use tweezers to handle them, and so far, that’s not working well…
thanks, Sixto
Last edited:
Thanks for the list! I will order some to try.
In the meantime, I hacked a temporary fix with a spare ZIF I had in my parts bin... cut off 4 pairs of legs I didn't need, to make it fit on the board. Bent and filed the remaining 4 pairs of legs, and it's a nice push-fit. There is room to get the transistors to touch, or keep them separate.
I was able to get 4 matched pairs from the preselected 117 HFE group on the left of the photo in less than 10 minutes. All 4 pairs have the collector voltages within +/- .1 mV which given the error potential of my DMM that's probably as close as I can get at this current level. I'm now using a 10V bipolar supply and the DIP switch set at 1.2k, and 620r (same readings reinforce my opinion of the accuracy of my relatively cheap DMM)
At higher currents, The 316r resistor gives a collector voltages difference of .4mV, and the 154r resistor makes the collector voltage difference 1.4 mV, but these are way higher currents than the actual circuit the transistors will be used for... any point in using these DIP settings to find closer matches?
In the meantime, I hacked a temporary fix with a spare ZIF I had in my parts bin... cut off 4 pairs of legs I didn't need, to make it fit on the board. Bent and filed the remaining 4 pairs of legs, and it's a nice push-fit. There is room to get the transistors to touch, or keep them separate.
I was able to get 4 matched pairs from the preselected 117 HFE group on the left of the photo in less than 10 minutes. All 4 pairs have the collector voltages within +/- .1 mV which given the error potential of my DMM that's probably as close as I can get at this current level. I'm now using a 10V bipolar supply and the DIP switch set at 1.2k, and 620r (same readings reinforce my opinion of the accuracy of my relatively cheap DMM)
At higher currents, The 316r resistor gives a collector voltages difference of .4mV, and the 154r resistor makes the collector voltage difference 1.4 mV, but these are way higher currents than the actual circuit the transistors will be used for... any point in using these DIP settings to find closer matches?
Attachments
Last edited:
I was going to ask ... what are all the capacitors for? Can you share the schematic? You only need to bypass the power supplies.
The nicer finish socket will allow the pins to be inserted much easier than the cheap one you have. I tried the cheap ones, threw them all out and bought good ones. This makes all the difference in the world!
Your meter is only used as a null device in this application. That relaxes accuracy concerns unless you want to figure things out. What I would highly recommend is a used HP meter, like a 34401A (I used those before the new ones), a 3457A would be excellent (make sure the internal battery is good and it is calibrated). A 3456A is good. All those meters are top of the line and you'll learn to love them! I have every meter I suggested, and most still on the bench. I've had other meters also, but if you're going top spend some money, get one of these I listed. Not unless you can afford the new current meters. Then get one of those (they have additional useful features), the trend function being valuable.
Just test at currents similar to what they will be used at. 1 and 2 mA is very common for tail current.
The nicer finish socket will allow the pins to be inserted much easier than the cheap one you have. I tried the cheap ones, threw them all out and bought good ones. This makes all the difference in the world!
Your meter is only used as a null device in this application. That relaxes accuracy concerns unless you want to figure things out. What I would highly recommend is a used HP meter, like a 34401A (I used those before the new ones), a 3457A would be excellent (make sure the internal battery is good and it is calibrated). A 3456A is good. All those meters are top of the line and you'll learn to love them! I have every meter I suggested, and most still on the bench. I've had other meters also, but if you're going top spend some money, get one of these I listed. Not unless you can afford the new current meters. Then get one of those (they have additional useful features), the trend function being valuable.
Just test at currents similar to what they will be used at. 1 and 2 mA is very common for tail current.
I’ll take them out and see what difference it makes. This is when I wish I knew more about circuit design and functionality, to understand the effect of keeping or taking out a component from a pre-existing design. I know just enough to get into trouble.
The film capacitors are parallel to the base and collector resistors, so perhaps they are providing a smoothing function to keep the voltage readings from jumping around too much? it doesn’t look like they would impact the collector voltage readings, but I matched the resistors so closely, and did not match the capacitors, (so that could throw things off.?) Maybe I can learn a little more by modeling the circuit in LTSpice and changing values around to see what happens…
So much to learn, so little time!
The film capacitors are parallel to the base and collector resistors, so perhaps they are providing a smoothing function to keep the voltage readings from jumping around too much? it doesn’t look like they would impact the collector voltage readings, but I matched the resistors so closely, and did not match the capacitors, (so that could throw things off.?) Maybe I can learn a little more by modeling the circuit in LTSpice and changing values around to see what happens…
So much to learn, so little time!
Given that it is my design, and the original has worked for over 30 years, you are in little danger of having any problems.
This design was distilled down to be as simple as possible while producing the desired results. The DUTs never did need bypass capacitors. There isn't gain enough to oscillate. I was more concerned with the current sources and simply good design practice.
It worked on perf board, and plug in experimenters boards. Generally speaking, if it works okay on those things, it is stable.
This design was distilled down to be as simple as possible while producing the desired results. The DUTs never did need bypass capacitors. There isn't gain enough to oscillate. I was more concerned with the current sources and simply good design practice.
It worked on perf board, and plug in experimenters boards. Generally speaking, if it works okay on those things, it is stable.
Good question.
My personal feeling is that parts that are not required should be removed. They may add unknowns to the mix. I know for a fact they are not required, just the power supply bypass capacitors should be installed. In fact, I used a good power supply and with the short leads I had even those bypass capacitors were not required. However, I put them in because I never know what someone else is going to use for power, or me in the future. Just good design practice.
When I originally saw the bypass caps I was baffled. We have seen the design become complicated beyond what is needed. But, that's fine. As long as the thing works properly, everything is fine. Adding the capacitors simply increased the cost of building it. For example, I could have made the current sources exact, and track between polarities - or variable on top of that. But when I asked myself what it did to produce better results, the answer was nothing.
The design goal was to create a way to exactly match transistors that wasn't available. It was to be as simple and as fast as possible while also performing at the highest level reasonably possible. Metering became simple as well, use the meter as a null device. Simple, direct, fast. Hopefully most meters with auto zero would be accurate in this application. The mV reading may be off, but as close to zero as you can get is what counts. Difference values then become a qualitative indication of your match if you want to figure it out in percentage. I retained the ability to measure base and collector currents (inferring emitter current) so you could figure out the beta of the parts. I don't anymore. I use a different instrument for that.
My personal feeling is that parts that are not required should be removed. They may add unknowns to the mix. I know for a fact they are not required, just the power supply bypass capacitors should be installed. In fact, I used a good power supply and with the short leads I had even those bypass capacitors were not required. However, I put them in because I never know what someone else is going to use for power, or me in the future. Just good design practice.
When I originally saw the bypass caps I was baffled. We have seen the design become complicated beyond what is needed. But, that's fine. As long as the thing works properly, everything is fine. Adding the capacitors simply increased the cost of building it. For example, I could have made the current sources exact, and track between polarities - or variable on top of that. But when I asked myself what it did to produce better results, the answer was nothing.
The design goal was to create a way to exactly match transistors that wasn't available. It was to be as simple and as fast as possible while also performing at the highest level reasonably possible. Metering became simple as well, use the meter as a null device. Simple, direct, fast. Hopefully most meters with auto zero would be accurate in this application. The mV reading may be off, but as close to zero as you can get is what counts. Difference values then become a qualitative indication of your match if you want to figure it out in percentage. I retained the ability to measure base and collector currents (inferring emitter current) so you could figure out the beta of the parts. I don't anymore. I use a different instrument for that.
I am going to ask a pretty dumb question but this is an area that I am not super familiar with so I will now proceed to make an knucklehead of myself.
Just using an example of the 2SC3478/1376’s that were listed earlier in the thread for testing, how does one go about sorting out the best 2SC3478’s too then be used in the matching process discussed with in the thread? I am assuming one would want to do this to get the best of each side of the set, or am I mistaken in this.
Thank you very much in advance for your time and help.
Just using an example of the 2SC3478/1376’s that were listed earlier in the thread for testing, how does one go about sorting out the best 2SC3478’s too then be used in the matching process discussed with in the thread? I am assuming one would want to do this to get the best of each side of the set, or am I mistaken in this.
Thank you very much in advance for your time and help.
Hi exojam,
First, you want devices that fall in the normal range for beta. Transistors that have higher or lower than the normal range of beta are either fakes, or they are abnormal. Plus the chance of finding matches is low. Best culled and used for very non-critical applications.
Pre-grade the beta ranges to save time. The closer these are to being the same temperature, the better the results you will have. Then use the matcher on parts the same range, and also one up and one down. Pair the matched parts. You can measure the beta of the pair using whatever meter you want, or calculate it from measurements off the beta matcher. If you have a complimentary diff pair (NPN pair and PNP pair), pick pairs close in beta. They will not match perfectly even if your beta numbers are the same.
Getting matches is the most important. Higher beta may matter a bit, but matching is by far more important. I'll go with tight matches every single time.
First, you want devices that fall in the normal range for beta. Transistors that have higher or lower than the normal range of beta are either fakes, or they are abnormal. Plus the chance of finding matches is low. Best culled and used for very non-critical applications.
Pre-grade the beta ranges to save time. The closer these are to being the same temperature, the better the results you will have. Then use the matcher on parts the same range, and also one up and one down. Pair the matched parts. You can measure the beta of the pair using whatever meter you want, or calculate it from measurements off the beta matcher. If you have a complimentary diff pair (NPN pair and PNP pair), pick pairs close in beta. They will not match perfectly even if your beta numbers are the same.
Getting matches is the most important. Higher beta may matter a bit, but matching is by far more important. I'll go with tight matches every single time.
Anatech,
Thank you very much for your time and explanation. So the main question that I would have after reading your response is how do you test the transistors to get them ready to be used in one of the circuit testers described and used in this thread? Is that another circuit board that somebody has created and is available on this forum, or do I just need to read up on how to place say 45 in a row and provide them the necessary voltages to test for proper output? I hope that makes sense as again I am not really familiar with testing transistors. Thank you again.
And I hope the grammar is correct since I do not have my glasses on.
Thank you very much for your time and explanation. So the main question that I would have after reading your response is how do you test the transistors to get them ready to be used in one of the circuit testers described and used in this thread? Is that another circuit board that somebody has created and is available on this forum, or do I just need to read up on how to place say 45 in a row and provide them the necessary voltages to test for proper output? I hope that makes sense as again I am not really familiar with testing transistors. Thank you again.
And I hope the grammar is correct since I do not have my glasses on.
lol!
No problem.
You can use the transistor beta check function on some cheap DMMs. They are very approximate and some may not give repeatable results. All you can do is avoid touching the transistors and leaving them in long enough to settle to room temperature. It's a real pain.
Bend the leads to fit sockets in advance to reduce handling them at the time. Use pliers or tweezers to handle them to reduce settling time. Once you have them graded, then use the tester outlined here. Tweezers or pliers again. The pairs have to settle thermally.
It is patience and time. Sorry, no reasonable shortcuts.
No problem.
You can use the transistor beta check function on some cheap DMMs. They are very approximate and some may not give repeatable results. All you can do is avoid touching the transistors and leaving them in long enough to settle to room temperature. It's a real pain.
Bend the leads to fit sockets in advance to reduce handling them at the time. Use pliers or tweezers to handle them to reduce settling time. Once you have them graded, then use the tester outlined here. Tweezers or pliers again. The pairs have to settle thermally.
It is patience and time. Sorry, no reasonable shortcuts.
