DIY Class A/B Amp The "Wolverine" build thread

Of course, the other reason was the failure to remove R17 and J103. Ouch. Won't make that mistake again
I don't know of any reason why leaving R17 and J103 connected would cause any problems.

Keeping the parallel 9.1K resistor with R17 only causes the bias through the leds to increase from ~1.3mA to ~6.7mA

Keeping J103 installed only causes Q107 and Q108 to run in class AB instead of class A.

You may need to investigate further as I don't think that was your issue.
 
After completing my Wolverine EF-4 as shown in this post I needed to create a nice high end preamp. It should have electronic controlled input selector, volume control and tone control including loudness switch. Furthermore 4 line inputs, a DAC and a bluetooth receiver and a MM/MC phono preamp.
That rather complex project is now almost finished:

Total.jpgOpen Top.jpgFront Open.jpg
I decided to use a ESP32-Devkit processor which has enough power to control all required function including web interface, infrared receiver and a SPI (serial) interface to connect displays, port extender, a serial RAM to store all settings (battery backed up).

The first build was using a relays controlled resistor network. Also the first tone controller was also a relays controlled resistor network in a baxandall circuit. This variant was not so perfect as expected. Slightly different switching times of those relays got sometimes to click noise.

So the final design uses now a MUSES72320 chip for volume control and a NJW1119A chip for tone control. Those chips also need SPI to control their functions. Relays are still used for the input selector, the mono and the loudness switch and also the Tape Monitor/Source switch. Two rotary encoder are used, right for volume/mute, left to control all other functions through menu structure and also to power on/off.

Each input has its own line amp with adjustable gain, I can set sensitivity from -10db to +6db. The ESP32 has also some A/D converters, which I use to emulate VU-meter which helps a lot to set correct gain on all inputs.

I developed a rather extensive web interface and integrated also a infrared remote receiver by software.
Here some screenshot of the provided web pages:
web1.PNGweb2.PNGWeb3.PNGWeb4.PNGWeb5.PNG
Works over WiFi on any modern browser. For infrared I integrated a old Apple TV remote.
 
Almost all the transistors failed. One cause I suspect is that I didn't have effective grease cover. I'd appreciate any views on this from the photo.

Of course, the other reason was the failure to remove R17 and J103. Ouch. Won't make that mistake again.
Yeah so Stuart has confirmed no harm done by leaving the temp R17 parallel resistor and J103 in. Thats good.

Your problem is still thermal in nature I think, these outputs were not making sufficient contact with the heatsink. This is not just due to "how tight the screws are" - this is due to the transistors not sitting flush with the heatsink, somehow. Either from the screw feeling tight but in fact bottoming out rather than torquing up on the transistor; or the screw being tight, but the transistor has been soldered in a bit short and is raising up somewhere; or the screws thread has been stripped and is not getting the correct torque to get flush to the heatsink. The two center ones would have definitely overheated and shorted IMO.

1730163748859.png
 
Yeah so Stuart has confirmed no harm done by leaving the temp R17 parallel resistor and J103 in. Thats good.

Your problem is still thermal in nature I think, these outputs were not making sufficient contact with the heatsink. This is not just due to "how tight the screws are" - this is due to the transistors not sitting flush with the heatsink, somehow. Either from the screw feeling tight but in fact bottoming out rather than torquing up on the transistor; or the screw being tight, but the transistor has been soldered in a bit short and is raising up somewhere; or the screws thread has been stripped and is not getting the correct torque to get flush to the heatsink. The two center ones would have definitely overheated and shorted IMO.

View attachment 1373864
Ok, I understand better what to avoid next time. Upon reflection I could see that I didn't get them sitting flush on the heat sink.

Do you mean the four circled are likely candidates for failure? Looks that way. Do the three in the middle look ok based on grease pattern.
 
Chiptech:

Shoulder washers are usually made of nylon or a similar plastic. The narrow end fits into the MOSFET and the wider end sits on top. Its purpose is to ensure that the bolt used to secure the MOSFET to the heatsink is centered in the MOSFET's mounting hole and does not come into contact with the MOSFET itself. RickRay is probably correct -- it is commonplace now for MOSFETs to have an insulated hole, obviating the need for a shoulder washer. I have plenty of uninsulated MOSFETs that require shoulder washers and wondered if that might be the case for you.

Regards,
Scott
 

Attachments

  • shopping.jpeg
    shopping.jpeg
    2.3 KB · Views: 34
@Chiptech, are all your heatsink tapped holes absolutely vertical? The circled mounting photos seems to show otherwise with the uneven goo distribution.
A drill press is strongly recommended and great care is needed for hand tapping. One of the reasons for using the large Belleville washer is that it can accommodate small amounts of offset angle but only when there are no other washers used AND when it is mounted the correct way up.
 
  • Like
Reactions: Mainframe