Adcom GFA-4702 rebuild

Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.
I thought I'd post about my current hybrid car/home amp project. I purchased this Adcom GFA-4702 several years back. It was missing the power supply chassis and cables but it was a good deal so I picked it up with the intent of building my own power supply. I sat in in my pile of projects 5yrs ago and I finally dug it out today to give it a test using a power supply I had out of a home amp.

The amp board was in nearly immaculate shape except for a repair someone had performed on a ground trace from the 8 pin molex where the front end regulated voltages are routed. It looks like somebody had miswired this connector in the past and toasted about 1 inch of the ground trace. Their repair left a lot to be desired and they covered it in silicone so it looked like hell. I removed all of the silicone and their attempt at a trace repair and repaired the trace properly. No silicone needed with this repair. :D

Initially I had planned to build up a new switching supply to use with this amp but once I realized I had an old supply out of a home amp in my parts bin that would work it made it hard to spend the $$$ on a switch mode supply.
My homebrew supply works out just about perfectly with both channels of the amp properly biased (60W draw per channel) the power supply sits at about +/- 29V for the unregulated supplies and +/- 33.5V for the regulated front end supply which is just about perfect.

I'm going to set this power supply up to have fuses on each DC rail in addition to the fuse on the AC mains. So it'll have a bit more protection than it did using the original power supply that Adcom used.

I'm driving it temporarily in the picture with an old MP3 player and it sounds great. I've got a friend that is going to strip and powdercoat the chassis for me and I've got the parts on order to build up some original wire harnesses. My plans are to purchase a medium sized chassis off of Ebay to put this power supply in and build up cables about 6 feet in length so the supply can be positioned on the floor with the amplifier up higher.

I'm not 100% certain what I'll use this amplifier for or if I'll even keep it in the end. I could use it as an ultimate garage amp or even try it out in my home stereo if the fan doesn't prove to be too obtrusive. I'd like to try it out on my Acoustats to see how it sounds driving a pair of electrostats. I'll leave the 4702 amplifier itself original (aside from some parts that need replacing) so if I sell it another user could couple it with an original Adcom power supply or drive it from my homebrew AC mains power supply I'm building.
 

Attachments

  • 4702_Internal.jpg
    4702_Internal.jpg
    436.5 KB · Views: 176
  • 4702_Test_Setup.jpg
    4702_Test_Setup.jpg
    414 KB · Views: 176
I started some refurbishment work on the 4702 yesterday. I'm not sure if this is a really early unit but it had some odd component screw-ups.

One of them is the capacitor C903 that is a 1000uF/16v and is installed across the fan. I noticed that the original one was a bit swollen on top. I thought it was probably due to years of heat exposure. I happened to have a cap to install in this location so I replaced it. I also replaced the feedback down capacitors C405/505 which are supposed to be 220uF/25v. The one in the "B" channel C505 was actually a 1000uF/16V and matched the brand and manufacture date of C903 mentioned earlier. This was not a big deal but it does shift the roll off much closer to DC for that channel. I also replaced the bias network caps C401/501 while I had it apart. After finishing these capacitor replacements up I powered that amp back up and after about 30min I noticed that the cap C903 I replaced was already showing signs of swelling on top again. I powered the amp down and flipped the board over and sure enough the cap C903 when installed per the silkscreening has the polarity reversed. I checked my pictures of the board before I started the repair work and the stock one was installed in the same orientation so this was wired up in reverse polarity all of these years.

The reversed polarity for C903 is ultimately due to the modification that the factory performed on all of these old Adcom's fan control circuit where they cut the traces on the board and P2P wired the cap into the circuit. I have a GFA-4302 and a 4304 available to me so I opened them up and they are wired correctly so it looks like only the 4702 was only affected.

In case you're wondering I'm sure these caps were factory installed as the solder was untouched and the date codes on the caps are as expected 1993.

I'm also replacing all of the trimmers. The gain trimmers are mangled so I'm ordering new single-turn trimmers for those and the Bias & offset trimmers I will replace with multi-turns to make adjustment easier.

Now I just need to get a case for the power supply on order.
 
Well its taken me a couple of months to get everything together for this 4702 power supply build. I sourced the PS case from China and that took a solid month to show up. I attached a picture of it after completion. I also sent the 4702 chassis out to a local professional powdercoating company and I just got it back the other day. It looks top notch now.

There was another fairly serious factory screw-up in the amplifier.
After finishing up the amplifier power supply once I started testing the amp I noticed that the fan was always running at full speed. A bit of searching and I found the PTC thermistor that controls the fan had one leg desoldered and lifted from the board. I reinserted the fan speed PTC back into the board and the fan speed dropped down to barely spinning and never spun up even when the amp got very very hot. I only had about 0.2V across that PTC. The resistor feeding the PTC (R904) had close to 12v across it. An in circuit resistance check showed R904 to be high. I started checking the resistor color codes and sure enough R904 was 7.5k and R903 was 2.7k. This is the opposite of what they should be. These resistors are right next to each other and I guess the assembler got the parts mixed up when the amp was built. I know these parts have never been touched as the solder was completely original.

This resistor swap was a serious screw-up which left the fan not spinning up as the heatsink temp increased and caused the overtemp remote shutdown to engage prematurely due to R903 being too low of a value. Its hard to believe that this could've left the factory like this and wasn't caught in some sort of quality testing. I guess it speaks poorly about Adcom's QC back then. Anyway if you have some of these old Adcom amps and there's a fan speed control problem check those resistor values.

I'm doing some testing on the amp now. My plans are to move it into my home system for a while to see how it sounds driving some electrostatic speakers.


4702_PS_front.jpg

4702_PS_rear.jpg
 
Sorry I should've been more clear on that. The PTC's in these amps are bonded with some good adhesive to the heatsink. The PTC itself was never removed and still has its original adhesive bonding it to the heatsink. Someone in the past had pulled one leg up out of the PCB (but left the PTC bonded to the sink) which drove the fan to max speed and I guess that solved their heat problems, albeit at the expense of full speed fan noise all the time. I just resoldered the PTC into the board and started troubleshooting for the real problem. Which led to finding the resistors R903 & R904 swapped.
 
Here's my quick THD tests @ 8ohms & 4ohms. I don't have a 2ohm rig as most of my work is on home audio and I don't feel like cobbling my loads into parallel to run a single channel 2 ohm test. Based on what I'm seeing its looking about as I expected. My 4 ohm power is a bit low but the rails under full load are a bit lower than expected at about +/- 26vdc.

THD_vs_Pwr_4ohm.jpg

THD_vs_Pwr_8ohm.jpg
 
After running this 4702 through my home system on various speakers I found another issue with this amp. When playing on my Acoustat's the amp when powered off would emit a bit of a low freq motorboat whizzing sound through the speakers. The left speaker only for a split second but the right speaker for about 2 secs. It didn't exhibit this behavior on 4 other sets of speakers, all of which were 4 - 8 ohm cones and not e-stat's. I started looking at the amp circuit and wondering where the sound would be originating from. My thoughts were that it had to be in the front end as the output stages take much longer to bleed down.

I put the amp on the bench and as I started looking at it I remembered a zobel network that was factory added to my Adcom 4302 & 4304 amps. This 4702 did not have the factory added zobel's on it. I pulled the covers off my spare 4302 and identified the added parts (2.7k resistor & 4.7/50v e-cap) and the location of the zobel (across zener D401/501). This zener diode is the reference for the front end current sources.

I dug out the required parts and added them to my 4702. I only had a 2.49k resistor, but that's close enough for this purpose. I reassembled the amp and now when powering the amp down with the Acoustat's connected there's nothing but silence. Just as it should be.... I had always wondered what the factory was trying to correct when they had tacked on these zobel networks, now I know.

I included pics of the 4302 zobel's and one channel of the 4702 with my newly added zobel. I also added a portion of the 4702 schematic that shows the location of the zobel parts in red in case someone needs to add them. The front end electronics in all these Adcom 4xxx series amps is identical so these diodes are easy to identify for any Adcom, be it a 4404, 4402, 4702, etc.

I still have a bit of a thump when the amp is powered on, but this is probably going to be left alone. In the SMPS from the factory they designed the SG3525 for the amps front end power supply to power up first and using a longer RC constant they bring the SG3525 for the output stage SMPS power up last. This works well to eliminate the power on thump. In my power supply here I don't have this option. I added a CRC filter to slow the charge of the output stage capacitors up a bit and this helped with the power on thump somewhat. My only other option that I can think of to eliminate the thump completely with my linear supply would probably involve relays and more large resistors to greatly slow the output stage charge cycle and I don't really want to make this power supply that complicated. I also do not really have the option to add an output muting/protection relay in the amplifier chassis.

4302_Zobel_a.jpg
4702_Zobel_a.jpg
Zobel.jpg
 
Status
This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.