Head-Fi.org › Forums › Misc.-Category Forums › DIY (Do-It-Yourself) Discussions › Matched transistors in Dynahi
New Posts  All Forums:Forum Nav:

Matched transistors in Dynahi - Page 3

post #31 of 158
11/19/04 at 11:46pm
Quote:
Originally Posted by randytsuch
I figured the output transistors, since they must act in concert with each other, would have a bigger effect if they are not matched, so I would like to match them as close as possible.
Actually, the output transistors have the benefit of global negative feedback to compensate for mismatchings. The input differential stages, and their constant current source transistors, on the other hand, are much more sensitive to mismatches because they are the very devices responsible for making global negative feedback work.
Reply
post #32 of 158
11/20/04 at 3:30am
Thread Starter 
Quote:
Originally Posted by amb
Actually, the output transistors have the benefit of global negative feedback to compensate for mismatchings. The input differential stages, and their constant current source transistors, on the other hand, are much more sensitive to mismatches because they are the very devices responsible for making global negative feedback work.
Oh great, now I am going to worry about how well the other transistors match up.

I was also wondering if I could test these other transistors, which are already installed on the board. If I apply a small positive dc voltage to the input (of course nothing will be connected to the output), and measure the outputs from the front end transistors. Then, I could reverse the input leads to apply a negative input voltage, and measure again.
Is this reasonable?

Thanks for any advice,
Randy
Reply
post #33 of 158
11/20/04 at 5:01pm
Quote:
Originally Posted by randytsuch
I was also wondering if I could test these other transistors, which are already installed on the board. If I apply a small positive dc voltage to the input (of course nothing will be connected to the output), and measure the outputs from the front end transistors. Then, I could reverse the input leads to apply a negative input voltage, and measure again.
Is this reasonable?
Randy
You might as well just power up the amp when you have everything except the DC servo installed, and measure the DC offset at the output. If it's excessive, then check the voltages at various points of the circuit to see if there is any obvious points of asymmetry. Don't forget to short the input to ground when you do this (if you don't have a volume pot connected). With a pot, set the volume to minimum position.
Reply
post #34 of 158
11/22/04 at 11:08am
The last devices didn't match so great, so I am ordering a few more:

2SC3381BL, hfe @ 1mA
Left sides:
310.3
427.2
575.6
corresponding right side:
291.3
696.0
575.6


2SA1349BL @ 1 mA
Left sides:
408.9
445.2
581.9
659.4
corresponding right side:
449.8
543.0
559.8
679.5

I couldn't get some of them down to 1 mA, so some measurements were done at 1.5 - 2.5 mA. Even 2.5M ohms wasn't enough in the Rb value to get the collector current down. I'll just assume matching within the same range is better than no matching at all... it at least let me catch one of the 2SC3381BL that was a dud.
Reply
post #35 of 158
11/22/04 at 4:30pm
Quote:
Originally Posted by chillysalsa
I couldn't get some of them down to 1 mA, so some measurements were done at 1.5 - 2.5 mA. Even 2.5M ohms wasn't enough in the Rb value to get the collector current down. I'll just assume matching within the same range is better than no matching at all... it at least let me catch one of the 2SC3381BL that was a dud.
Yes, that's true.
Reply
post #36 of 158
11/22/04 at 6:09pm
Quote:
Originally Posted by chillysalsa
hFE = (mA METER) / (Irb)
where Irb = VRb / Rb while Rb is measured with the power off.
Quote:
Originally Posted by Pars
I constructed a circuit similar to what was posted. Since I did not have any suitable pots (nor did ratshack), I decided to use a fixed resistor of 220K, giving an Ic value of 10-11mA on the PNPs (2SA1015). Since I know the resistance up front from measuring it, and powering it with 9V from a bench supply, I can more or less assume that when testing in one batch (i.e., one sitting without powering the supply down) that the base current (Ib) will be constant. I will still measure Vb and compute the base current each time unless I see that Vb is holding steady (it seems to be). I will obviously measure Ic for each device and compute the hFE from the measurements.
Quote:
Originally Posted by amb
The input differential stages, and their constant current source transistors, on the other hand, are much more sensitive to mismatches because they are the very devices responsible for making global negative feedback work.

Bear with me please, I'm no electronics sage Putting together those quotes above, I understand the critical matching seems to go for 1145 and 2705 (input differential stages? Or do the 1349 and 3381's also belong there?). So in any case, how about the following ultra-cheapo-simple transistor matching procedure:


1) Fixed Rb, let's say 220K like Pars did

2) Supply is just a 9V rectangular little battery

3) I first measure VRb as follows:


4) Then I measure mA meter as in the image that has been posted already:


5) Then I compute that transistor's hFE = (mA Meter) / (VRb / Rb)


After doing that for a bunch of them, I can pick the 1145's with closest hFE's to the corresponding 2705's. Maybe same for 1349's and 3381's.


Would that be at least better than no matching, or would that ultra-cheapo-simplistic-naive transistor matching procedure be really useless?
Reply
post #37 of 158
11/22/04 at 7:08pm
Yeah, I think that's right (I was matching for a dynalo, so without the dynahi schematic in front of me, I am just guessing that the tranny #s you gave are correct).

Amb's suggestion was to match based upon position in the amp at a representative current (Ic) for that position (amb, correct me if I am wrong). So he advocated a pot so you could dial in the resistance (Rb) to give you te appropriate Ic for that amp stage, and then measure the Vrb and Ic for each, calculating the hFE based on that.

Since I didn't know the exact Ic for each position in a dynalo, I just used ~10mA as a close enough guess (tho some are at 1mA, some at 15mA) to do my measurements at. One transistor I did play around with resistors to give ~1.6mA and ~10mA. The hFEs calc'd out to 240 and 236, respectively, so they weren't moving around alot for changes in Ic (at least for that one).

The procedure I used was to let the transistor settle briefly, and using two meters at once, read both Vrb and Ic at approximately the same time. Marked each transistor, recorded the numbers and used a spreadsheet to calc hFE when I was done measuring. As had been noted, as the transistor heats up, it will change, so consistency is probably important as far as when you take the measurement. I guess ideally you might also measure transistor temperature?
Reply
post #38 of 158
11/22/04 at 7:18pm
I see Pars. Well, I only have one meter, could get another similar one since it was just about $15 if I remember well. However I don't have some generic test bench 9V power supply. That's why I would like to use just a simple 9V DC battery despite the low current. I can certainly attach the transistors to a heatsink, but not sure it will heat up much just with such a battery.

My logic goes that if two transistors are really well matched, they should measure almost identically even in my simple test, right?. Now the reverse I'm not so sure of, which is what I would like to know. If two transistors measuring almost identically in my simple test might still be really off in higher current and temperature conditions. But if they are off already in this simple test, well then those would be bad to use for sure I guess.

So the simple test might provide some degree of satisfactory matching. Wonder if that's the case even if done with just one meter powering on and off between measuring voltage and mA. I think I'll get another meter then.
Reply
post #39 of 158
11/22/04 at 7:27pm
About the Dynahi schematic:
http://members.cruzio.com/~billsiegmund/D2r3.pdf
Reply
post #40 of 158
11/22/04 at 11:04pm
Quote:
Originally Posted by Pars
Since I didn't know the exact Ic for each position in a dynalo, I just used ~10mA as a close enough guess
My calculations of the collector currents of the BJTs in the dynalo are: 2mA each for the CCS transistors, 4.3mA for the VAS stage, and about 18mA for each of the output transistors.
Reply
post #41 of 158
11/22/04 at 11:20pm
Quote:
Originally Posted by amb
My calculations of the collector currents of the BJTs in the dynalo are: 2mA each for the CCS transistors, 4.3mA for the VAS stage, and about 18mA for each of the output transistors.
Thanks amb. One of these days (years?) I'm going to remember/relearn everything I haven't used in 20 years so I can figure those out myself
Reply
post #42 of 158
11/29/04 at 11:15pm
Thread Starter 
So, I have assembled both boards now, and tested them.
Without the pots or opamps, one has 14 mv of offset , but the other has about 90mv of offset .

Now, I need to look at the second one, and see what is causing that big offset.

Randy
Reply
post #43 of 158
11/30/04 at 2:12pm
Quote:
Originally Posted by randytsuch
So, I have assembled both boards now, and tested them.
Without the pots or opamps, one has 14 mv of offset , but the other has about 90mv of offset .

Now, I need to look at the second one, and see what is causing that big offset.

Randy
Without any transistor matching, and also without the pots or opamps, one of my boards had -16mv of offset, the other +34mv.
Reply
post #44 of 158
12/5/04 at 10:01pm
Looks like MCM is out of stock on the Toshiba 2SC3425. Does anyone have another distributor to recommend? Been matching and baggin transistors all day. I want some more parts for tighter matches.
Reply
post #45 of 158
12/5/04 at 10:11pm
Thread Starter 
Quote:
Originally Posted by dgardner
Looks like MCM is out of stock on the Toshiba 2SC3425. Does anyone have another distributor to recommend? Been matching and baggin transistors all day. I want some more parts for tighter matches.
Did you mean the 2SC3421?
bdent has them.

Randy
Reply
New Posts  All Forums:Forum Nav:
  Return Home
Head-Fi.org › Forums › Misc.-Category Forums › DIY (Do-It-Yourself) Discussions › Matched transistors in Dynahi