A cautionary tale

I decided to post this as a word of caution to people looking at second hand / DIY products
- especially if you do not have the skillset ( or access to someone who does ) to repair / troubleshoot.

“Built and purchased directly from well known DIY amp builder, xxx xxxxxx. USD $600.00

Firstwatt Aleph J circuit. Dual mono power supply using Hammond toroid's, 4x18000uf panasonic caps per channel, CRC configuration. Dead silent operation, largely due to the isolated grounds.
Runs hot, but you need Aleph's to run toasty for the liquid 3D sound character of single ended class A. Unobtanium toshiba 2sj74 input jfets, matched quad (gain match and minimize any offset)”

https://www.usaudiomart.com/details/649958166-first-watt-aleph-j-class-a-clone/images/4346437/







On visual appearance – not bad, a little industrial but that suites class A DIY. The description looks solid with all the right keywords – bolded.

Retrospectively, it would have been good to ask for something to gauge size / aspect and to open the top to see the internals.

DOA on arrival but that turned out to be a simple disconnection of one of the primaries from the IEC , front panel was buckled and top was pushed in – these are both exceptionally soft material . Front plate looks like some sort of sound damping material used in automotive , thin alu front and back with a sort of foam sandwiched inbetween. Top perforated sheet is 1mm , flimsy.





First things noted – no IEC ground – so chassis is not grounded. “isolated grounds” was a bolt wrapped in gorilla tape with the center tap, amp ground and psu ground connected together. The advertised 4x18,000uF capacitance. When I build supplies like this, I use 33,000uF caps with 2 per rail in CRC. So what we have is a typical FirstWatt amp power supply that has been modified for dual mono , hence cutting capacitance per rail in ½ . Final first observation was that there were no supports for the PCB, the mosfets were holding the board in place (not terrible but not optimal) .


Seeing as the AJ is capable of a differential input - improves noise in amp and the “chassis” had holes cut for XLRs, I added XLRs with a switch for SE and Balanced input. I also added full IEC ground and chassis ground with a lift in case RD experienced noise in his system relating to ground loops.



I added a proper ground , keeping them separate and adding the standard FW use of a varistor to isolate the chassis/IEC from the circuit.



Initial test showed massive variation between sides and between individual mosfets, as an example I found one FET running at 2amps while another was at 600mA bias – this equated to one mosfet running at close to 90deg C and the heatsinks sitting at 85deg C ( for those who have no reference – this is $#%^&$ hot ) and the fets would not last long under these conditions.

I measured the bias across each fet and given there was such variability even after I added a trimmer to be able to adjust the bias down to be more manageable, that I had to pull the amp boards and replace the mosfets which were clearly NOT a matched quad .

Not unexpectedly – it sounded off, dull and some hint of distortion.



part 1.

 

part 2.

Luckily I did this as I discovered a few fun things

1. A short between a resistor lead and another resistor.



And a number of solder joints which appeared to be poorly soldered – I suspect that this was soldered only from the top of the board – again not terrible, not optimal BUT if you do this, you have to use a hot iron and make sure the solder flows through completely. This was not the case – the amp worked, so there was some connection between all the necessary elements to make it work.



Post reflow – all joints well soldered



A few final pics of the rebuild – tapped supports for the PCB, new ceramic thermal coupling with a good dose of heat paste.






Now working and sounding as it should.




not to steer people away from great deals , but be aware that you might be buying a can of worms

..dB

 

Did fixing the shorts/bad solder joints in the amp boards allow you to use the existing FET's or were they still unmatched?

 

no , I tested the pulls and after testing a few had confirmed enough variation to explain previous behavior. So I replaced the fets with a properly matched quad per side - matched at 24V under 1.2A , so there is no question that they will be matched under load. I used IRFP140 which are now unobtainium and have better spec than the IRFP240 that are current production. It is good to have a parts bin that has all sorts of bits stashed away for rainy days.

 

I googled the name of the "well known amp builder", and it didn't take long to find a 13 year old thread on Audiogon with harsh warnings. How these numpties keep getting away with selling crappy gear never ceases to amaze me.....

 

It always amazes me how capable these First Watt clones are - despite this one being limited on total power ( not enough heatsink to crank up the bias ) , it still sounds incredible and able to extract some beautiful sound from my maggies ( not concert level , but plenty for casual listening )



Balanced operation and SE work as anticipated , absolutely silent on both my test bed speakers in the basement and the maggies.

A pic to give perspective for how big the heatsinks should have been for a 25W classA amp - I have dialed this one back to about 18 W



This AJ is still highly capable but bias is much more appropriate for the heatsinks - the psu and rest of the circuit is capable of being dialed up, the transformers are Hammond and provide a solid 23.5V dual rails after rectification and under load.

Even though this arrived as a firehazard, it is an amp I would happily own with the few changes made. It is now a true dual mono Class A FirstWatt Clone that is stable and will likely keep kicking for at least 10 years ( the expected lifespan for caps run in such a hot environment ) . The cosmetic touches would be easy to address - a custom faceplate and top would make this look the part but it would be just as functional left with it's current adorning.

..dB

 

Looks great now, man! Glad it was salvageable.