Yet Another Adcom GFA-565 Thread

Administrator
Joined 2004
Paid Member
Hi tsip,
Many thanks. I'm glad the forum has been helpful for you.

You have the only factory installed fan circuit I have ever heard of. You have a rare beast all right! Do you hear the fan running, or does it track the heat linearly?

-Chris
 
I was going to mount the fan above the control board blowing down, onto the board and into the box. I realize this is working against convection, but it is the only way to really assure the components on that board are being cooled.

So my plan is to mount two small (1.5x1.5x.6) 12 volt fans on standoffs just under the cooling slots on the center housing (the u-shaped housing in the center rear, between the heatsinks. This housing holds the control board, as well as the various I/O connectors.

Per Anatech's suggestion, I got a small solid state thermostat, and a regulator, so I'll regulate the amp DC down to 12 volts and then run the fans with a series resistor, thereby lowering their speed. The resistor will be shunted by the normally open thermostat, so when the temp gets high, the fans will blow at full speed. Simple and easy...
 
Hi tsip,
Many thanks. I'm glad the forum has been helpful for you.

You have the only factory installed fan circuit I have ever heard of. You have a rare beast all right! Do you hear the fan running, or does it track the heat linearly?

-Chris

And thats not the only unique thing about my GFA-585. It has been rebadged as Dynaudio!
I enquired with Dynaudio and the story is that they imported some in Europe under the then pro branch of the company. By the looks of it, the amp has been probably seating for years in some studio until it blew up some monitors and they got rid of it. I ll post some pics one of these days..
To answer your question, no, the fan has never kicked in as my speakers cannot be driven that hard to make this monster sweat to that point.

Again many thanks Sir. You are a real asset to this forum.

Cheers
 
Administrator
Joined 2004
Paid Member
Hi tsip,
Amazing, but it makes sense too. They probably designed as part of that contract and made it an option should something like that ever be needed. It wasn't, but now what you have said explains that little mystery. Back in the day, my service contacts at Adcom were emphatic that they hadn't supplied any amplifier with the fan option. I always wondered why a never installed option made it into the service manual.

Thanks, Chris.
 
Administrator
Joined 2004
Paid Member
Hi Scott,
Why not simplify it further and use 120 VAC fans?

Many 12 VDC fans have a built in speed regulator that will attempt to maintain the set fan speed. There are some that have two speed settings, and many others for computers that have a tachometer output. You may well run into a speed regulated type of 12 VDC fan. If you have a variable power supply, try the fan out before attempting to set it up for two speed operation.

Fans that run on AC usually do not have speed regulation.

Please let us know how this works out for you. I'm interested.

Best, Chris.
 
I built up the test circuit per your schematics, and re-tested the NPN Darlingtons.

Here are the results. You can see that the numbers ARE rather different. The really out of bounds transistor (#6) was still out of bounds, and in general the pairs that were close in the last test were close in this test, although the specific matches changed a bit.

The collector voltage differences in this test were higher, and, since I was using 0.1% resistors, the offset voltages were much lower.

Thanks Anatech for your guidance on this !!

NPNMatches2.jpg


I will set up the PNP test today and post those results later. With 10 devices (9 in this last test, since I seem to have lost one!!) this test takes a long time. I found that each device needs about 2-3 minutes to settle to a stable value, and I had to cover the test setup with a small plastic cup to keep the air currents from changing the measurements.
 
Last edited:
Administrator
Joined 2004
Paid Member
Hi Scott,
Yes, your time is about right between tests. Try leaving a close pair in overnight and check them in the morning. If your foam gives good isolation and there is a cover, the pairs will settle down more quickly. I find there is a slight thermal oscillation, sometimes more pronounced than others.

You must have done your single measurements very carefully. That also takes time. Anyway, I'm really happy you find that method helpful. Once I built one up, I began using it constantly. It became obvious that my breadboard creation was being used to death, so I committed the design to copper.

-Chris
 
My first measurements were also in pairs. I test the pairs both ways in the circuit: For example #1 in position A and #2 in position B, and then reverse them. Doing this, you can back out any differences in the differential circuit. It doubles the amount of testing, but it gives very good result.

I too saw that thermal oscillation. Often the pair will settle down over about a 2 minute period, and then slightly overshoot the min or max voltage difference, band then settle back by about 0.2 to 0.4 Mv.

I left one pair in while I was off doing something else, and after 4 hours they had drifted about 0.2 mV form what I had measured. So I think the measurements are pretty close.. certainly within a mV or so.
 
Well, I buttoned it all up today, and applied power to the control board, and poof.. Zapped the negative supply 220 uF cap. I removed the cap and briefly applied power to the circuit and measured -86 volts at that cap. Given that this is a 35 volt cap, something is obviously wrong.

By applying power I also found that R154, the 1.2K ohm resistor in the negative bias circuit was also smoked.

Not sure about R154, but the only way I could get -86 volts on the cap would be if the zeners D117 and 119 were either open or in backwards. One was open.

I also found that the other cap was very hot, and found that the voltage on that one was +86 v. Those zeners seem to be OK.

Got any ideas?

Scott
 
Last edited:
Administrator
Joined 2004
Paid Member
Hi Scott,
Everything sounds bang on for your matching.

The series resistors would have to be way low to cause that. An open zener sounds like defective from new. They tend to short when they die. Check the other zeners, then double check the values for series resistance.

-Chris
 
Noticed that R153 was also hot. Both of them check out at 1.2 K.

I also checked R155 and R156 which both show 27K ohms, So that all seems right.

It seems I have two problems. One is the improper voltage at the caps, and the other is the heavy current on R154 and 153.

I am going to work on the improper cap voltage issue first.

I think a start would be to isolate that part of the circuit, and determine either why the zeners are not doing their job, or what is otherwise causing this issue.
 
Administrator
Joined 2004
Paid Member
Hi Scott,
Either the zeners are open, or they aren't connected in circuit. It can't be anything else. They are a shunt circuit element that fails short. The only way they can fail open is if the body is blasted apart.

You might want to check in case they are regular diodes marked as a zener. That would be nasty.

-Chris
 
Yeah, I was thinking the same thing looking at the circuit. if the series resistance was too low, then they might pull too much current and pop, but that would be pretty obvious

They are the LM329's from Linear. I am going to pull them out and put them in a test circuit to verify that they break down properly.
 
OK, I tested the zeners and found two shorted and one open.

I put them back into the circuit, and tested them with the 27 K ohm resistor to the supply out. I did this using a 3.3K ohm resistor connected to a power supply, to verify that above 14 volts at the supply the voltage at the dies was fixed at 14 volts. It was. Both sides. I them put the caps back in, and applied power...still wight he 27K ohm resistors out of the circuit.

One of the positive supply zeners instantly popped. Since this was not yet connected to the positive supply, this tells me I have a short somewhere in the input stage. Looking at the diagram, if the base collector junction of Q107 is shorted, then I would get raw 86 volts at the zener, which would explain a lot..

So, I am off to pull Q107 to see if that's the issue
 
I figured out my problem...

I was using an MPSA 42 as an SC3478 replacement. These are pretty much the same electrically HOWEVER, the packages are different. The MPSA 42 uses a conventional EBC (left to right from the bottom) pinout, while the SC3478 uses the less typical ECB pinout. I didn't bother to check for that..

The result was I was connecting the base to the supply. I suspect this simply applied the full supply voltage to the base circuit that leads directly to the zeners..

This is also the case in the other area, where i was smoking R153 and R154...

Crap..

Scott
 
Administrator
Joined 2004
Paid Member
Hi Scott,
That isn't good. I have to say that I've never run into anything like this before. Let's hope you don't run into any more strange problems.

Those zener diodes are low noise LM329 I think. Replaced many due to electrolyte leakage destroying them.

-Chris
 
Administrator
Joined 2004
Paid Member
Hi Scott,
That will do it every single time. Sorry to hear that you learned the hard way.

Replace all the LM329's, and maybe try and get either the originals or a Japanese substitution. A TO-126 case transistor will go in without twisting the leads around. Just turn the package around and you're good. But do check the pin out every time.

-Chris
 
I twisted the leads on the MPSA transistors, and installed them, double checked everything (had fried two of the LM329s), and tested it once more before applying power.

No smoke!!

Progress!!!

I think the LM329's were bad from the start, and that's why some joker cut the traces to the caps (to keep them from exploding).. silly...

Tomorrow I'll start tracing the signal path to see what's going on there..

Thanks for your help!!

Scott
 
Last edited: