RMI-FC100, a single stage audio power amplifier

Too bad that ONsemi f..ked up some production lots of NJLs. I'm convinced that they put by mistake other type of diode inside which requires other compensation current in order to correct track the transistor Vbe over temperature.
I do not recommend to bypass all the diodes string with the pot. Just use what Andrew recommend, one resistor over the diode with way out Vf than the others.

Seems odd that they would mistake the diode type, but I also have experienced some problems. In my instance, it seems more a case of variable behavior.

One of the two amps I built would suddenly bias high, then lower. I chased the problem down to one diode that was apparently unstable. It would measure slightly different than the others with the diode setting on the multi meter, but would show about 50% higher voltage drop under load. I had originally bypassed another diode, and when I opened that connection that one read over 1V!! on the MM. However, when I ran current through that one, it dropped back to the normal reading, and was normal with the FC current load.

The take home lesson: Measure all your diodes with the expected operating current (about 2mA). Measure each one a couple of times. I would then mount them on the heat sink and short one with a clip lead. Adjust bias, then check bias with each one shorted before you button things up. If you do not shut the amp off when you move the clip lead, you should short the next one in line before you remove the clip on the prior one - or your bias will be way too high.

Sheldon
 
Sheldon, I have bought 175 of each NJL0281DG and NJL0302DG for the German FC-100 group-buy from VERICAL some weeks ago.
A DIY-friend of mine matched them and also measured the Vf of the integrated TT-diodes at a current of approx. 2.5mA. (12V / 4K7).
He did all the measurements twice.

The Vf of the NJL0281 - TT-diodes is in the range of 509 - 518mV; the Vf of the NJL0302 - TT - diodes is in the of 519 - 529mV.

Some of the Germans have already built the FC-100.
Their builds work with 5 active TT-diodes, and I have not heard about any adjustment / TEMPCO problems.

I therefore think that if you buy the NJLs from the current production, you won't have any problems with the internal TT-diodes.

Best regards - Rudi
 
I therefore think that if you buy the NJLs from the current production, you won't have any problems with the internal TT-diodes.

Thanks Rudi. My comments were more directed to any users who might have Thermaltraks already on hand. I think I have solved my issues by just checking each diode under actual operating conditions, and selecting the best one to bypass. Had I known that there might be some variability, I'd have checked more carefully before I installed them. Both my amps (four channels) are fine with 5 diodes in the string.

Sheldon
 
It's been quite a while since I became interested in this project and bought the boards and other parts in the first group buy. Sourcing parts was particularly frustrating; but finally, after many interuptions it is working as the mid drivers in my Linkwitz LX521 dipole speaker system.

I have been of the opinion that perceived differences between half decent power amplifiers would be miniscule: particularly where my 70 year old, and none-too-golden ears are concerned. How wrong I was. Particularly in the spatial domain these power amps have improved the system as much as going from the Orion speakers to the LX521's. It becomes aparent that linearity is hugely important in uncluttering the Acoustic Scene. Presumably intermod' products generate a lot of spatial "fuzz". As others have remarked, the amps seem to have total control of the drivers from low frequencies to the prickliest/sharpest transients. One of the first things I noticed was that I became very aware of the inner voices/parts such as the violas in string ensembles; indicating that some masking sounds were absent compared to lesser amplifiers. On well recorded material the AS is more simple and believable than anything I have heard before.

I almost had the amps going in December last year, but stopped work to build the LX521's. Unfortunately when selecting output devices for the second channel and finding the best place to put them on the heatsink I had some moments of madness that did not help progress. While removing and installing the board for the nth time suddenly one of the PNP thermal traks is hogging all the current. Assuming that it was faulty or that my numbering system was defective I wrongly concluded that it would have to be unsoldered. In the event I cut the leads off and removed the remaining pieces one at a time. Lo and behold a similar thing happened on the NPN side of the amplifier and this device was removed intact with a little bit of damage to the pad at the top of one via/thru hole. I now realise that I had forgotten to finally tighten the screws through the OP devices into the heatsink:mad: Loose screws =bad thermal bond=high Iq

As I have the slower thermal tracks I applied a 1UF film capacitor between input ground and the chassis. The first attempt was unsucessful at preventing oscillation (C across input connectors) but moving the caps directly between the input pin and the nearby back panel via shot leads did the trick.

Mihai, I can understand why you rank this design as the best you have heard. Congratulations, excellent work.

Keith
 
I happened to read this thread only yesterday. This design seems perfect to drive fast rising broadband speakers. But alas component woes by the number. Thermal track BJTs unavailable and the sK170 only with the GR spec. If I could get some thermal tracks via ebay , could I get along with just 1 pair in the power stage? ( Then a Vbe multiplier is necessary of course)
 
I happened to read this thread only yesterday. This design seems perfect to drive fast rising broadband speakers. But alas component woes by the number. Thermal track BJTs unavailable and the sK170 only with the GR spec. If I could get some thermal tracks via ebay , could I get along with just 1 pair in the power stage? ( Then a Vbe multiplier is necessary of course)

I can sell you mine if you want,very likly I shall never build my FC100
 
C8, C9

Is it possible to increase the value of either C8 or C9 (both 15pf silver mica caps) on the main board?
The reason I ask is that I mistakenly ordered one 15pf and one 18pf instead of two 15pf. Would using these two be detrimental to the cct design or should it be OK?
If OK, which one should be 18pf; C8 or C9?

Thanks
 
Is it possible to increase the value of either C8 or C9 (both 15pf silver mica caps) on the main board?
The reason I ask is that I mistakenly ordered one 15pf and one 18pf instead of two 15pf. Would using these two be detrimental to the cct design or should it be OK?
If OK, which one should be 18pf; C8 or C9?

Thanks

The difference is to small to cont. Don't worry ;-)
 
Transistor Matching

Hi Mihai! I am in the process of constructing your amp FC100 and do not quite understand hoe to match output devices - NPN-PNP by hfe - i.e. NPN-PNP - 1 = hfe 1; NPN-PNP - 2 = hfe 2 etc., and the same with vbe? Or, all the devices must have the same vbe and hfe? Please explain. Thank you.

U$1 Q2 NJL0281D Must have the same Vbe +-1mV with the same NPN or PNP with matched hfe​
U$2 Q12 NJL0281D Must have the same Vbe +-1mV with the same NPN or PNP with matched hfe​
U$3 Q25 NJL0281D Must have the same Vbe +-1mV with the same NPN or PNP with matched hfe​
U$5 Q4 NJL0302D Must have the same Vbe +-1mV with the same NPN or PNP with matched hfe​
U$6 Q5 NJL0302D Must have the same Vbe +-1mV with the same NPN or PNP with matched hfe​
U$7 Q27 NJL0302D Must have the same Vbe +-1mV with the same NPN or PNP with matched hfe
 
Hi Mihai! I am in the process of constructing your amp FC100 and do not quite understand hoe to match output devices - NPN-PNP by hfe - i.e. NPN-PNP - 1 = hfe 1; NPN-PNP - 2 = hfe 2 etc., and the same with vbe? Or, all the devices must have the same vbe and hfe?

YMMV, but in general you will not find both the NPN and PNP with the same Vbe at similar operating conditions - in the case of the NJL/NJW pair above, they will defer by 10s of mV. Also, you will generally find the PNP beta to be higher than the NPN beta at similar operating conditions.