Adcom 555 and leaky caps

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
Administrator
Joined 2004
Paid Member
Hi Phloodpants,
Yup. You can easily check the state of the filter caps with an oscilloscope. I also change them upon request, but if they are okay I'll contact the customer to let them know. You are spending their money after all.

I leave things neat with proper lead dress, but going to great lengths to create new buss bars might be a little excessive since it won't be seen.

Input transistor match is critical as you know. How do you match yours?
My primary concern when I restore these amps is reliability. Every component and every solder connection is checked. I make no outlandish claims about improved sound quality, and I am very cautious about second-guessing the designer.
That's a healthy attitude, one I support completely. It's amazing how many techs don't understand how an amplifier works.

-Chris
 
Yep, I scope the filter caps, at idle and under load. Most are fine. I check em with the ESR meter too, but it doesn't ever seem to show any issues, even when they don’t scope well.

I match the input transistors with a cheap DMM with hFe testing. The tester runs at 1ma, and the Adcoms run them from .5mA to 1mA, so it seems a reasonable method. (Could be wrong.) But I get low DC offsets on Mk1 Adcoms this way, and on MKII's I check the output of the servo for excessive correction.
 
Administrator
Joined 2004
Paid Member
Hi Phloodpants,
I match the input transistors with a cheap DMM with hFe testing.
I was afraid you would say that. In order to measure transistors accurately, you have to do one of two things. You either must keep the transistors at exactly the same temperature (and I do mean exactly), or you can cancel out the case temperature when matching.

Build yourself a diff pair circuit where the DUTs take the place of the two transistors. No feedback, but do use a current source. Mine is adjustable. Clamp the cases together and shield them from external air currents. Measure across the two collector load resistors - differential style. You can use the DMM to presort your transistors, then go for the balance act to get parts that actually do match closely.

-Chris
 
...........I match the input transistors with a cheap DMM with hFe testing. The tester runs at 1ma, ................
Most cheap DMM set to hFE operate with a near constant Ib, not Ic.

The meter monitors the collector load voltage.
eg.
device A
Ib~1µA, hFE = 100, collector current ~ 100uA, meter reading 100, Pq ~0.3mW
device B
Ib~1µA, hFE = 300, collector current ~ 300uA, meter reading 300, Pq~0.9mW

But the big failing is that the Ib depends on Vbe. A high Vbe device will run at lower Ib than a low Vbe device. i.e. the hFE could be the same but the meter reading would not take account of the changed Ib resulting from the different Vbe.
 
Last edited:
Administrator
Joined 2004
Paid Member
Hi Andrew,
It is pointless to match based on Vbe. Additionally, you really need to keep the devices within an extremely close temperature, or make the temperature term fall out of the method. By placing the two transistors in physical contact and isolating them from the local environment, you can keep the two parts the same temperature. The temperature term now cancels out when the devices are matched. Select on the basis of identical voltage drop across collector resistors and you have now matched the beta and compensated for Vbe differences too. By using the differential pair circuit, you also replicate how most matched devices will be used.

Vbe differences tend to be within a couple mV between devices. If the beta is matched, the actual offset created by a Vbe differential will be .... you guessed it, a couple mV. Matching beta is the critical parameter that you want matched.

Hi Phloodpants,
Use 100 R resistors, closely matched. Stick about 10 K from the bases to common. I ordered 0.1% resistors from Digikey for these four parts. Double that if you are making the opposite polarity as well (I did, and I am using both right now).

You can use an LED based CCS for tail current with select-able current levels (the emitter resistance in the LED CCS). You can then have the freedom to set the tail current near the level the devices will be used. Normally you would be 1 ~ 2 mA, but at the moment I am using 10 mA for this particular amplifier.

Don't match with unequal collector loads as you will not have matched parts. The engineer who designed that amplifier made the assumption that the diff pair would be matched perfectly (we all do). Therefore, create a perfect match. The application needs to have the differential collector loads for something other than throwing the match off.

You must allow the pair to stabilize thermally. Because you have already grouped the parts by Hfe with the meter, your pairings with this jig will be reduced a fair amount. The extra wait time can be passed by doing other chores and checking on the match occasionally. If you have good isolation, the pair will converge more quickly. It's going to be at least 15 min for them to settle in. The meter looks at the differential voltage between the collectors. If yours can store readings in a Max-Min-average, your job will be easier as you leave it go and look at the average. You have to wait before starting the averaging unit the transistors are varying slowly back and forth.

Matches produced this way will be extremely tight, so much so that you will need to match degeneration resistors, or they will throw your match off. Yes, that close.

Happy jig building. I've had mine for years now.

-Chris
 
Hi Andrew,
It is pointless to match based on Vbe. Additionally, you really need to keep the devices within an extremely close temperature, or make the temperature term fall out of the method. By placing the two transistors in physical contact and isolating them from the local environment, you can keep the two parts the same temperature. The temperature term now cancels out when the devices are matched. Select on the basis of identical voltage drop across collector resistors and you have now matched the beta and compensated for Vbe differences too. By using the differential pair circuit, you also replicate how most matched devices will be used.

Vbe differences tend to be within a couple mV between devices. If the beta is matched, the actual offset created by a Vbe differential will be .... you guessed it, a couple mV. Matching beta is the critical parameter that you want matched.
...............
Paralleled devices require Vbe matching.
 
Administrator
Joined 2004
Paid Member
No they do not!
The emitter resistors force current sharing at higher currents where it matters. Matching the beta is the way to go for outputs as well as any other transistor position.

Think about it for a moment. How much variation would you expect in Vbe between like devices? Any drop across an emitter resistor will swamp out variations in Vbe easily. In fact, tolerance variations with the emitter resistors can throw off a matched set of transistors. On Semi has indicated a THD reduction is the result of having transistors with matching beta. Matched transistors will also have close Vbe characteristics anyway, so by looking at Vbe, you are completely ignoring the important characteristic. By matching beta, Vbe differences will be small.

-Chris
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.