Well, this has been instructive. Here I was thinking I had improved one of Nelson Pass' designs by putting the input buffer JFETs close enough together to thermally couple them.
Of course I had to turn them 90º to do that, leaving one between the heatsink and the other one. I'm pretty sure now that such a change would be losing ground, so it's back to the drawing board.
Anyway, thanks for the education, folks.
Of course I had to turn them 90º to do that, leaving one between the heatsink and the other one. I'm pretty sure now that such a change would be losing ground, so it's back to the drawing board.
Anyway, thanks for the education, folks.
I'm experimenting with immersing the DUT in a beaker of (magnetically stirred) distilled water. Distilled water seems to be quite a poor conductor of electricity, both in theory and in practice. It's cheap to buy and easy to discard, unlike various oils.
Try this little experiment at home: connect a 6.5 digit bench DVM to a 1 megohm resistor and set to "OHMS". Note the meter reading (a) before immersing in distilled water; (b) while immersed. Calculate the equivalent parallel resistance of the water.
Try this little experiment at home: connect a 6.5 digit bench DVM to a 1 megohm resistor and set to "OHMS". Note the meter reading (a) before immersing in distilled water; (b) while immersed. Calculate the equivalent parallel resistance of the water.
Hi Mark,
Did that experiment. Pure H2O is an extremely poor conductor of electrons.
Unfortunately, at the air-water interface, metallic objects introduce enough contamination (after a long while) to upset the non-conducting water. Any voltage difference between metallic objects can then begin to plate one with the other.
Coke on the other hand is just plain nasty stuff, should anyone be curious enough to want to make the leads disappear.
-Chris
Did that experiment. Pure H2O is an extremely poor conductor of electrons.
Unfortunately, at the air-water interface, metallic objects introduce enough contamination (after a long while) to upset the non-conducting water. Any voltage difference between metallic objects can then begin to plate one with the other.
Coke on the other hand is just plain nasty stuff, should anyone be curious enough to want to make the leads disappear.
-Chris
Hi Phloodpants,
In practical application, there simply isn't room for a nice fuzzy layer and metallic cap. However, if you experiment with closed cell foam you might find something that improves the situation. I have seen plastic caps that are attached to circuit boards that encase a transistor to prevent air currents from creating low frequency noise. This is an effective fix for that problem.
I'm wondering if a cap can be found that will encase the diff pair without being too large. You would still grease them and hold them together somehow (heat shrink still gets my vote), then allow the air trapped in with the transistors buffer the temperature difference. This might make a difference in phono and Mic amplifiers. Amplifiers tend to be okay with grease and heat shrink.
You've done a lot of hard work there, and that is to be commended.
Best, Chris
In practical application, there simply isn't room for a nice fuzzy layer and metallic cap. However, if you experiment with closed cell foam you might find something that improves the situation. I have seen plastic caps that are attached to circuit boards that encase a transistor to prevent air currents from creating low frequency noise. This is an effective fix for that problem.
I'm wondering if a cap can be found that will encase the diff pair without being too large. You would still grease them and hold them together somehow (heat shrink still gets my vote), then allow the air trapped in with the transistors buffer the temperature difference. This might make a difference in phono and Mic amplifiers. Amplifiers tend to be okay with grease and heat shrink.
You've done a lot of hard work there, and that is to be commended.
Best, Chris
Fortunately, distilled water is inexpensive (USD 0.89 per 3.7 liters). Unlike mineral oil or transformer oil or transmission oil, distilled water is so cheap you can throw it away after a single use, and NEVER have to store it.
How about using warm air? One could make an insulated glove box, from an old cooler and a big plexiglass window on top, and fitted with some kind of heater and thermostat. The fan mounted to the board seems to hold temperatures evenly, though too cold at room temperature. If the air around it were hot, I think it would work!
"Because it would be cool", "just for the hell of it", and "to see what it's like" are all valid reasons in my book.
This is DIY -- there's no requirement for it to be pragmatic.
Hi Jeff,
That is true, however Phloodpants has a commercial interest in both the process and the equipment required for matching. The time it requires to match these parts must be reflected in the selling price on the invoice to the end user - the guy who owns an Adcom amplifier.
Hi Phloodpants,
I've been matching transistors for Adcom amplifiers even before we became authorized warranty for them. It does make a difference to sound quality. Walt Jung has also done work in this direction in order to improve the performance. He has had some success with his approach. Unless he has released the information on his site, I can't discuss it.
One reason why I designed the matcher was due to the transistor types used in some Adcom amplifiers (and a couple others). They didn't lend themselves to beta measurements so another way had to be found. That resulted in the matcher design.
-Chris
That is true, however Phloodpants has a commercial interest in both the process and the equipment required for matching. The time it requires to match these parts must be reflected in the selling price on the invoice to the end user - the guy who owns an Adcom amplifier.
Hi Phloodpants,
I've been matching transistors for Adcom amplifiers even before we became authorized warranty for them. It does make a difference to sound quality. Walt Jung has also done work in this direction in order to improve the performance. He has had some success with his approach. Unless he has released the information on his site, I can't discuss it.
One reason why I designed the matcher was due to the transistor types used in some Adcom amplifiers (and a couple others). They didn't lend themselves to beta measurements so another way had to be found. That resulted in the matcher design.
-Chris
I recently purchased a beta matcher PCB from Chris Hoppe to help me get my hands on the input transistors I need for my Adcom GFA555II rebuild. I am aware that he does business repairing and upgrading these and other Adcom amps, but I wanted to do this project myself. The time I’ve spent building, using, and understanding the beta matching hardware and methods has been very interesting and educational to me. I’ve been happy to offer back some (I hope) helpful information about all of this through my posts and PM’s. (I’m deep into the matching process and will have more thoughts to share soon.)
A link to “Hoppe’s Brain”, Chris’s company, is shown as a signature at the bottom of all Phloodpants posts, so it shouldn’t come as a surprise to anyone that he’s in the business and has a financial interest in improvements to the transistor matching process. He offers up his board designs on the website, has sold some of his matcher boards, and shares his ideas for improving the hardware, so I’m completely happy to share any of my ideas and/or improvements with him or anyone else, whatever their 'gains' may be.
I think this thread is an excellent example of the benefits that can come from a community forum such as DIYAudio.
A link to “Hoppe’s Brain”, Chris’s company, is shown as a signature at the bottom of all Phloodpants posts, so it shouldn’t come as a surprise to anyone that he’s in the business and has a financial interest in improvements to the transistor matching process. He offers up his board designs on the website, has sold some of his matcher boards, and shares his ideas for improving the hardware, so I’m completely happy to share any of my ideas and/or improvements with him or anyone else, whatever their 'gains' may be.
I think this thread is an excellent example of the benefits that can come from a community forum such as DIYAudio.
Thanks Mike! My business grew out if my lifelong DIY audio interests, I am trying my best to respect the line between DIY and commerce, and always open to suggestion.
For the record, for the few people I've sold matcher boards too, I only charged enough to cover expenses on parts and shipping, as it's not my design to sell.
For the record, for the few people I've sold matcher boards too, I only charged enough to cover expenses on parts and shipping, as it's not my design to sell.
Hi Chris,
I haven't got any problems with it. Feel free, and I do appreciate that you are helping the community here.
What's amusing is that I also repair audio equipment as a business. I don't have the energy to maintain a web site and do all the other things I'm doing.
You have my blessing Chris
-Chris (too many Chris' in the electronics business)
I haven't got any problems with it. Feel free, and I do appreciate that you are helping the community here.
What's amusing is that I also repair audio equipment as a business. I don't have the energy to maintain a web site and do all the other things I'm doing.
You have my blessing Chris
-Chris (too many Chris' in the electronics business)
Okay, now back to work. I've run several dozen pairs of transistors through my rig and come up with two decent (but not great) matched pairs. I've got 80 pieces to work with and I use a simple transistor tester (Atlas DCA55) to do a preliminary sort by beta. Pairs from each beta sub group get pre-tested on the beta matcher rig. In this pre-check, I’m just covering the two with a cap as a quick check to see if they are within a few mV of each other. Those that are close get tested again using my aluminum thermal cap (once you use it a bit, it’s actually pretty convenient).
Here's my issue: for any pair that shows less than 1mV C-C voltage, I swap them and do another test. There is always a difference and that difference varies quite a bit. Either Cogeniac or Phloodpants posted a while back about the relatively constant offset he found in his test rig. The differences I’m seeing are far from constant, varying between 1 to 4 mV. The couple of matched pairs I’ve come up with so far happened to have small “swapping differences” - both swapped positions still fell under 1 mV.
Any thoughts why this is happening?
Here's my issue: for any pair that shows less than 1mV C-C voltage, I swap them and do another test. There is always a difference and that difference varies quite a bit. Either Cogeniac or Phloodpants posted a while back about the relatively constant offset he found in his test rig. The differences I’m seeing are far from constant, varying between 1 to 4 mV. The couple of matched pairs I’ve come up with so far happened to have small “swapping differences” - both swapped positions still fell under 1 mV.
Any thoughts why this is happening?
Some Test Results for Diego Mike's 2N5551's
For what it's worth, I first sort the transistors according to their beta as measured with the aforementioned Atlas DCA55. For pairs that show the same beta value, e.g., 179 and 179 (most of these 2N5111’s fall between 160 and 190), I tend to see worst case final test mV differences of about 2 to 3 mV.
On those occasions where I’ve tested pairs with noticeably different betas, I tend to see about 2 mV per beta unit. So, two transistors with betas of say, 170 and 185, will show C-C mV’s of around 25 to 30 mV difference. I'm sure these behaviors are fairly specific to this particular transistor models and perhaps to some extent to my particular test rig and methods.
For what it's worth, I first sort the transistors according to their beta as measured with the aforementioned Atlas DCA55. For pairs that show the same beta value, e.g., 179 and 179 (most of these 2N5111’s fall between 160 and 190), I tend to see worst case final test mV differences of about 2 to 3 mV.
On those occasions where I’ve tested pairs with noticeably different betas, I tend to see about 2 mV per beta unit. So, two transistors with betas of say, 170 and 185, will show C-C mV’s of around 25 to 30 mV difference. I'm sure these behaviors are fairly specific to this particular transistor models and perhaps to some extent to my particular test rig and methods.