Matching Transistors

[jheu02]

I just received some replacement parts and wanted to match them as best as possible. I have a cheap DMM that has sockets for transistors and also an hfe setting. I also have a better DMM, but would only be able to do a diode test on them and check the voltages.

Which is the better way to match them?

Thanks,

John

 

[BinaryMike]

If I was limited to this equipment, I would do both. And hope the test current is somewhere near what the devices see in the intended application. You could get more reliable results by lashing up a test circuit to roughly simulate the application, and using your meters to measure DC voltage and current in the test fixture. The next step up would be to apply an AC test signal and measure raw AC gain.

 

[EchoWars]

DC gain matching with a digital meter is plenty good enough for small signal transistors in most applications, such as matching for a diff-input pair.

 

[avionic]

Just HFE match them..

Damn to slow again..

 

[jheu02]

Thanks all. Since I have the capability to do both tests, I'll probably go that route, as it will be a learning experience.

John

 

[P.L.F.]

For power transistors matching - more info here: http://www.audiokarma.org/forums/showthread.php?t=293316&page=8
Regards

 

[jheu02]

Just HFE match them..

Damn to slow again..

Well, I tried that but all I got was 000 and every once-in-awhile, a (-) would flash. I had verified b c e on the pins in the sockets and correct npn vs pnp so I guess my cheap meter is just that.

DC gain matching with a digital meter is plenty good enough for small signal transistors in most applications, such as matching for a diff-input pair.

Let's see if I can explain my next question: as to the matching, is it between channels (i.e. the pnp of R ch matches the pnp of L ch, same for npn?)

or

is it trying to match the gain of the pnp to the npn in each channel?

or

do you try to get all matching as closely as possible in both channels?

I've got plently of matching PNPs and NPNs, but they don't match each other, though some are close. I figure I'll do both channels, even though the Qs of only one channel are suspect (haven't tested them OOC, but I'm getting some matching b,c b,e readings and even some c,e matches).

John

 

[BinaryMike]

The reason for matching power output stage transistors is to minimize distortion in that stage. Left/right channel matching doesn't matter, if channel gain matching is the goal. That's entirely determined by the feedback network in a normally functioning amplifier. So you want to match a PNP/NPN set for each channel of the amp, but it's almost useless to select matched pairs based on their gain at idle current levels. You need measurements at several amps, which calls for a curve tracer with high-current pulse testing capability. A home-brew test fixture with a base drive pulse generator and a scope to measure collector current could do the job, if you're sufficiently determined. If not, then it's best to just accept the transistors you got. Most amplifiers don't depend on matching to meet their performance specs.

 

[ecluser]

DC gain matching with a digital meter is plenty good enough for small signal transistors in most applications, such as matching for a diff-input pair.

Yes, and there is a good reason for this. In most DMM, the base current in hFE test mode is close to 10uA. With typical hFEs, this gives a collector current in the range of 1mA to 10mA and most differential stages work with collector currents close to this range.

John, sometimes you may need to bent the leads in a gentle "z" shape, to improve the contact in the transistor sockets, particularily if the sockets were previously forced by large leads from medium power transistors.

 

[jheu02]

Well, I matched my transistors as best I could, actually got the hfe test to work. It was a matter of giggling the leads to get them to make good contact, so thanks for suggesting to bend them. Anywho, I now will need to reset the idle current. I've done so before on amps and receivers where they either have dedicated pins, or where it's been referenced to ground and one leg of a resistor.

The receiver I'm working on now states that it needs to be done on a "cold" unit, but I have a question as to the wording of where to attach the leads.

"connect a DC VTVM across R541 and R543 in the right channel output. Adjust R517..."

"connect a DC VTVM across R542 and R 544 in the left channel output. Adjust R518..."

From that wording which lead should go where on the resistors? Does it matter which gets the + and which gets the - ? And which leg of the resistors would I attach to, that which is closest to the output transistors, i.e. the ones tied to the emitters?

Thanks,

John

 

[EchoWars]

Would help a lot if we knew WTF it is you're working on. :scratch2:

 

[avionic]

Would help a lot if we knew WTF it is you're working on. :scratch2:

Scientific Wild Ass Guess... a Yamaha CA-1000..

 

[jheu02]

Scientific Wild Ass Guess... a Yamaha CA-1000..

No. My CA-800 is still giving me fits, but that's not what I'm working on now.

Would help a lot if we knew WTF it is you're working on. :scratch2:

Philips Labs AH-7951. I'd read some posts about connecting meters to the emitter resistors, but I haven't seen specifically which lead of the resistors and/or polarity. So I thought it was a generalized procedure if specific jumpers or pins weren't used.

Each channel has 2 TO-3 output transistors. For the R channel, R541 is connected to the emitter of one of these, R543 to the collector of the other.

The L channel, R542 to the emitter of one, R544 to the collector of the other.

The SM just states to connect the meter across the two resistors in each channel, and then adjust the trimmer to obtain a reading of 20mV +/- 5, but doesn't state which meter lead connects to which resistor, nor does it state on which lead of the resistor to connect to, that tied to the transistors or the lead away from them. I can post either that section of the schematic or the upper or lower board layouts if that would help, but I'll need to scan them first.

John

 

[avionic]

It doesn't matter,just adjust for 15 -25 mv..connect the test leads to the emitter resistor leads that are directly connected to the output transistors emitters. Look at the picture of the sample circuit.In this circuit you would connect your leads to TP3 and TP4 to measure your bias .

 

[jheu02]

Cool, thanks. Like i said, I thought it would be a fairly "generic" procedure, but all the amps I've checked and set bias on have had either the test pins installed on the board already, or have stated to connect the negative lead to the chassis or other ground point, so just wanted to make sure I didn't screw anything up.

John

 

[ecluser]

"connect a DC VTVM across R541 and R543 in the right channel output. Adjust R517..."

"connect a DC VTVM across R542 and R 544 in the left channel output. Adjust R518..."

In my opinion this means that you should connect the negative probe on the collector of the "lower" power transistor (Q517 in one channel, Q518 on the other channel), and connect the positive probe on the emitter of the "upper" power transistor (Q519 or Q520)

So, you must have 20mV across the two 0.47 Ohm resistors connected in series, for a bias current of ~20mA.

 

[jheu02]

I got the idle current set no problem and then checked DC Offset. L is reading ~20mV and R is up around 53mV. There's no adjustment trimmers for DC offset, so now I need to track down why it's so high. I should probably take this back to my original thread on the AH-7951, but should I be looking at resistors as possible culprits or diodes? :scratch2:

John

 

[avionic]

Actually the offset is not that bad. You maybe able to lower the offset by replacing the differential pairs in the amplifier(s) front ends with closely Hfe matched replacment transistors.

 

[jheu02]

Actually the offset is not that bad. You maybe able to lower the offset by replacing the differential pairs in the amplifier(s) front ends with closely Hfe matched replacment transistors.

Thanks for that schematic. Mine has a very similar layout, so it's nice now knowing what some of the terminology is and relating it to what it looks like on a schematic.

When comparing voltages on a meter to what's in the SM, is it correct to assume that, if the voltages are "close" to what's published, the difference could be chalked up to differing gain levels in transistors, or could it indicate issues with resistors or diodes being out of spec? I'm noting that some of the transistor voltages are listed as 0V, yet I'm seeing 30-60mV on some of them. Could that be related to the DC offset readings I'm seeing? Others are low by 1-2 volts in places.

I'm trying to get rid of a loud turn-off pop (turn-on is lower but still noticeable), and I can see the voltage spike on my DMM registering if I shut off the receiver with the meter attached to the speaker terminals. Goes from the earlier mentioned numbers, to the 250-300mV range. It's a combination Power/Speaker selector switch.

 

[avionic]

is it correct to assume that, if the voltages are "close" to what's published, the difference could be chalked up to differing gain levels in transistors

Assume nothing! Yes it could be as you said,It could be a typo on the schematic or transistor leakage-resistors or capacitors--