Wow, guys 
Thank you for so useful info.
Specially for polymer caps. Goood catch, I totally forgot for such option.
I hope there is no problem with so low DC rating.
As for R24 - hmmm, there no so much space for 1W resistor (or must mounted vertically).
OK, I will try to find... mouser ( or farnell and digikey) are my best friend.
anatech: Yes, transistors are my other battle... What I have now is NJL's, 2SC/SA and SK170 matched.
Seller tell me that batch is from GB, but with multi-meter I see some differences. Huh, I must use jig to prove that.
If something with NJL's are not OK, bad...
I need to go with new batch.
So, remain a lot of transistors to be matched.
Anyway, I think Roender's design deserve this careful choice and building.
I'm fan from long time ago
Kiril
Thank you for so useful info. Specially for polymer caps. Goood catch, I totally forgot for such option.
I hope there is no problem with so low DC rating.
As for R24 - hmmm, there no so much space for 1W resistor (or must mounted vertically).
OK, I will try to find... mouser ( or farnell and digikey) are my best friend.
anatech: Yes, transistors are my other battle... What I have now is NJL's, 2SC/SA and SK170 matched.
Seller tell me that batch is from GB, but with multi-meter I see some differences. Huh, I must use jig to prove that.
If something with NJL's are not OK, bad...
I need to go with new batch.
So, remain a lot of transistors to be matched.
Anyway, I think Roender's design deserve this careful choice and building.
I'm fan from long time ago
Kiril
Hi Kiril,
I designed a beta balancer for matching transistor pairs a while ago (actually many years ago). There are a few threads that you can find with a more or less complete schematic. Some members have designed PCBs for it. Look for an Adcom GFA-565 thread. One of those detailed the construction. There is a PNP and NPN section as well on the same PCB.
You will have to buy 0.1% resistors for the collector and base series resistance. I think the collectors are 100R0 and the bases were 10K0. The jig uses a constant current source for the tail current. It's very stable. Don't use a fan as some members have.
-Chris
Edit: I forgot to mention. Do not use Chinese terminals for the test sockets. They don't accept leads from even new transistors very well. I bought some to try as I figured Ebay would be the place most people would turn to for these. Buy them through proper distribution chains and you won't have any problems. I used Digikey for most of the parts.
I designed a beta balancer for matching transistor pairs a while ago (actually many years ago). There are a few threads that you can find with a more or less complete schematic. Some members have designed PCBs for it. Look for an Adcom GFA-565 thread. One of those detailed the construction. There is a PNP and NPN section as well on the same PCB.
You will have to buy 0.1% resistors for the collector and base series resistance. I think the collectors are 100R0 and the bases were 10K0. The jig uses a constant current source for the tail current. It's very stable. Don't use a fan as some members have.
-Chris
Edit: I forgot to mention. Do not use Chinese terminals for the test sockets. They don't accept leads from even new transistors very well. I bought some to try as I figured Ebay would be the place most people would turn to for these. Buy them through proper distribution chains and you won't have any problems. I used Digikey for most of the parts.
Hey Roender, I have the original FC-100 kit from the group buy of a few years ago, which I plan to build soon (finally).
Have been made some changes in the meantime?
Last edited:
Telstar, by my "novice" in understanding amplifier, is totally pointless to look for small gain in folded cascode design.
Healthy, wide bandwidth and High gain in only one stage are one of main advantage in this topology!
So, may be you must look in different direction if gain in FC-100 are so high for you.
Or... think about input transformer coupling, like me
with high turn ratio.4:1 for example.
But, this is very complicated and expensive area to get good results easy...
Telstar, by my "novice" in understanding amplifier, is totally pointless to look for small gain in folded cascode design.
Healthy, wide bandwidth and High gain in only one stage are one of main advantage in this topology!
So, may be you must look in different direction if gain in FC-100 are so high for you.
Or... think about input transformer coupling, like me
4:1 for example.
But, this is very complicated and expensive area to get good results easy...
I'll see what I can do. A good friend of mine pointed me to study a texbook where there could be some tricks to do the job without messing with the NFB.
I go DAC direct and I want to avoid anything inbetween. I'm trying an amp with the same amount of gain and I have to attenuate between 20 and 40dB, therefore I know what would be optimal.
Still, having the beautiful kit that Roender made, it's a real pity not to build it.
Anatech's Beta Matcher
Here is one version, I think:
http://www.diyaudio.com/forums/solid-state/157682-dc-adcom-gfa-555-a-60.html#post2158681
Here is one version, I think:
http://www.diyaudio.com/forums/solid-state/157682-dc-adcom-gfa-555-a-60.html#post2158681
Thank you for your help PH104! It's a bit difficult for me to find exactly this post in a big theme!
Finally I know how exactly look... Very clever to measure only difference
If all other things are kept equal, the phase margin and stability will primarily be determined by the amount of global feedback.
Therefore, if the open loop gain can be reduced (for example, by adding source resistors to the input JFETs) and the global NFB resistor re-chosen for lower closed loop gain, it should be possible to keep the total amount of NFB similar to the standard RMI-FC100 configuration, hopefully negating the need for other alterations to the circuit.
Therefore, if the open loop gain can be reduced (for example, by adding source resistors to the input JFETs) and the global NFB resistor re-chosen for lower closed loop gain, it should be possible to keep the total amount of NFB similar to the standard RMI-FC100 configuration, hopefully negating the need for other alterations to the circuit.
...long time...not time. First of all: Thx for your answer.
Reasons:
1. I have some parts which don't fit to the PCB. (Power Resistors with tighter tolerances but different package).
2. The main Idea behind it: P2P wiring, dead bug, ugly style or what so ever called supposed to have (if correctly applied) lower parasitics. PCBs are only an approach to make modern electronics cheaper and easier to maintain. But even with a "poorly" soldered p2p circuit you might reach a working state up to ghz range. Especially if you have a closed groundplane.
I vent change the positioning of the parts in the circuit. More or less I added only the 3dimension. Denser layout but with less crosstalk. Usually signal lines cross in a 90Deg angle. So less parallel wires. Input stage is still seperated to poweramp section but has now a groundplane. In the past Ive used this approach even for delicate clock circuits (smd) with very good results (also on the scope).
3. I can use the cu clad as "whole" heatsink for the low power transistors /diodes. The parts where glued together with 2k-thermalgrease and then attached to the clad (no change in the Thermocoupling scheme/layout - just a "cu-clad" substrate to hold them in place)
But here is the culprit: Yes if I want't to have it sounding like urs I should build it like yours. Which I didn't - so not equal performance is one of the consequences I have to face
..thats why I ask those questions to find the issues in my build.If it comes to EMI /Groundloops a cu-clad with free wiring should be superior to a dense doublesided pcb. The p2p layouts have less parasitics and so the circuit should behave more ideal.
At the moment I'm still suspecting an NFB issue. Or some local oscilation. Hower If someone has had similiar Issues like Oscilations etc. I would be really glad to get your input. Any sugesstions are welcome.
regards
Alex
Good Question. And also interesting for those of use who want to use (modern)available parts. The original 2sk170l is indeed nice but difficult to source.
I think compensation caps should be enough if you have similar problem. No need to change bias resistors. Btw, I have used diy pcbs (laser printer method) with very compact layout, perhaps this contributed to stability of this amp and lower output capacitance of op devices which shifted phase margin
...so according to the mkII schematic it musst be R6/R41. Should be equivalent to R6/R41 in this FC 100 version. By Compensation Cap you mean a Cap in Parallel to R6 respectively R41. Or from driver base to ground as a local low pass filter to catch high freq transients? (Does my thought make even sense)
I am revisiting my beloved FC100 - I want to replace by home made single sided PCBs with something better. For that purpose I created a compact layout, without big output caps and fuses which will sit on separate board. I would appreciate some feedback. Were there any changes in the schematics during the years?
