One could track down some japanese e gain group ones, in a batch of 1000 e group ones I purchased in SA about 7 years ago I have several that go above 700, measured on fluke meter,not so many to match though. I cant find more anywhere , luckily I have been able to get semi direct from sanyo the 2sc2362 group h trannies, equally excellent with gains above 700 with a lot of matches, it costs one pays for this though but still cheaper than zetex. They reserve these components for manufacturers, with sanyo parts Rotel comes to mind and one has to pull some strings to get them. It helps to have high gain for the offsets but I wouldnt compromise noise and other factors for it.
Western semiconductor technology is inferior and lags behind Im afraid, its good to see some like onsemi doing very good work in the higher current domain but their stepping stone was also to copy the japanese. One often sees western design, japanese manufactured even with ICs, not only because of favourable pricing but performance too. Sadly the japanese are moving away from the manufacturing of discrete devices, I surely hope that we will not be stuck with having to go backwards to bc and 2n parts or even to greater horror gainclone type circuits in future.
I hope the gainclone fraternity are not reading this.
Western semiconductor technology is inferior and lags behind Im afraid, its good to see some like onsemi doing very good work in the higher current domain but their stepping stone was also to copy the japanese. One often sees western design, japanese manufactured even with ICs, not only because of favourable pricing but performance too. Sadly the japanese are moving away from the manufacturing of discrete devices, I surely hope that we will not be stuck with having to go backwards to bc and 2n parts or even to greater horror gainclone type circuits in future.
I hope the gainclone fraternity are not reading this.
homemodder said:
I hope the gainclone fraternity are not reading this.
I've used chipamps and don't feel particularly dirty...
You are simply in love with the Japanese transistors, aren't you?
You are aware that this all was invented here, right?
If you were nearby, I'd challenge you to a test - you can tell me which of the identical amps has the ZTX696B as the differential pair. THAT would impress me.
Hi Sandyk
I built some lm type low power amps for my dad which used them for his ham radio interests when I was about 11 years old. The last time I used a gainclone was when I was 18, old new car, off to varsity and my car radio didnt play loud enough, didnt pay much attention to quality then, only loudness, but it was only 20 watt modules. Didnt know how to build good smps then to couple to serious amps. Too bad I didnt have something like this frugalamp which looks good indeed coupled to a smps.
I build amps mostly for car use, much more profitable, there is very little high quality ones on the market and they still a big hit with the youngsters.
I built some lm type low power amps for my dad which used them for his ham radio interests when I was about 11 years old. The last time I used a gainclone was when I was 18, old new car, off to varsity and my car radio didnt play loud enough, didnt pay much attention to quality then, only loudness, but it was only 20 watt modules. Didnt know how to build good smps then to couple to serious amps. Too bad I didnt have something like this frugalamp which looks good indeed coupled to a smps.
I build amps mostly for car use, much more profitable, there is very little high quality ones on the market and they still a big hit with the youngsters.
MJL21193, I would take you up on that, but with a headphone amp using my seinheisers, I doudt I would get one single test wrong. You could build one with the zetex and one with 2362. It would be unfair if I had to point the zetex out but I could point the lower cob part out, irrespective of the one you may use. With power amps its much harder to distinguish and I would most probably get some wrong, would be better if I could shave a couple of years off too. The very low cob trannies seem to have a softer and less agressive tone in comparison to higher cob parts, more like jfet sound when used in front end LTPs.
I'm not one to say what a person can or cannot hear, whether it's real or imagined. I do know that I, and many more beside me, would not hear the difference.
Sticking with the purely technical reasons to choose a particular device is probably better and less contentious.
Let's compare:
I see the 696B has a Cob of 6pF, yet it's input capacitance is 200pF. Is this high? Looking at the 2SC1845 I see that it has a listed Cob of 1.6pF, but they make no mention of the input capacitance. Calling up the spec sheet on the BC550 I see Cob = 3.5pF and Cib = 9pF.
The problem:
I have 70V rails. I have on hand several 2SC1845E but they would not survive my voltage! I'd prefer to cascode, but that would mean changing OS's design (oh the humanity!).
Conclusion:
I see the ZTX696B as a viable solution. I and many others will not detect the inferior quality of this device when used in this application. But that's always been my problem - I listen to the music, not the transistors.
Sticking with the purely technical reasons to choose a particular device is probably better and less contentious.
Let's compare:
I see the 696B has a Cob of 6pF, yet it's input capacitance is 200pF. Is this high? Looking at the 2SC1845 I see that it has a listed Cob of 1.6pF, but they make no mention of the input capacitance. Calling up the spec sheet on the BC550 I see Cob = 3.5pF and Cib = 9pF.
The problem:
I have 70V rails. I have on hand several 2SC1845E but they would not survive my voltage! I'd prefer to cascode, but that would mean changing OS's design (oh the humanity!).
Conclusion:
I see the ZTX696B as a viable solution. I and many others will not detect the inferior quality of this device when used in this application. But that's always been my problem - I listen to the music, not the transistors.
homemodder,
Way behind. Pronouncedly switch-mode technology, carrying its distinguishing features, not very exciting for audio, which demands a different approach, where basically the necessary knowledge is missing.
Just copying.
Superior manufacturing technology as well.
A truly terrifying perspective.
go ahead ,do the cascode, (only on the condition you use a zener
)About voltage..??? KSA1845's do`180v , since fa2 's LTP
only goes from -.8v to V+ rail (60-100v), I see no problem here.
Hearing the trannies
I do hear the difference in OP device's(mj15003's vs these new njw0281/0302's) as the new ones
sound brighter/clearer especially at high volumes.
Never A/B'ed LTP choices..
OS
Heres yer' cascode..sims great, shaves a little off the gain
as well.
Does not seem like a "frugal" anymore
.
OS
as well.
An externally hosted image should be here but it was not working when we last tested it.
Does not seem like a "frugal" anymore
.OS
MJL21193, I also listen to music, but I prefer the music to sound as close as possible to what I would hear on a stage. Compromises will always have to be made somewhere, sometimes I compromise for example in diamond buffers the output transistors by using higher cob devices to get higher 2nd harmonic distortion than could be achieved with better trannies for the quest of it sound more realistic or to improve a cold, thin and metallic sounding source.
Cascoding is a route one can take, or for 70v supplies you could find 2sd756a hitachis, very difficult nowadays. I for one agree with Pass sometimes less is more when it comes to complexity. Chasing ultimate THD numbers is sometimes counterproductive but good for marketing figures.
Os now youve gone and complicated a very nice design
If you chasing figures you could go futher and use cfps for the ltp and taking a step further bootstrap the front end, dont know whether youll like the sonic results though. To reduce gain you could split the vas ltp like the cyrus amp and boost the currents with subsequint EFs. It works well for THD figures and getting kick *** bass.
Lumba if you have any new ideas for smps, I would be all ears. Send me a Email. I ve gone bruteforce with sonic benefits for some customers where price, weight and the loss of boot space was of no importance by using massive 24 Volt truck batteries in series and in parralel. This also requires the addition of custom made alternators. The alternators themselves seem to be a problem, they generate noise too.
Cascoding is a route one can take, or for 70v supplies you could find 2sd756a hitachis, very difficult nowadays. I for one agree with Pass sometimes less is more when it comes to complexity. Chasing ultimate THD numbers is sometimes counterproductive but good for marketing figures.
Os now youve gone and complicated a very nice design
If you chasing figures you could go futher and use cfps for the ltp and taking a step further bootstrap the front end, dont know whether youll like the sonic results though. To reduce gain you could split the vas ltp like the cyrus amp and boost the currents with subsequint EFs. It works well for THD figures and getting kick *** bass. Lumba if you have any new ideas for smps, I would be all ears. Send me a Email. I ve gone bruteforce with sonic benefits for some customers where price, weight and the loss of boot space was of no importance by using massive 24 Volt truck batteries in series and in parralel. This also requires the addition of custom made alternators. The alternators themselves seem to be a problem, they generate noise too.
Hey OS,
How about doing a constant power cascode?
That is, string a 2mA CCS from the positive rail to a resistor of 10K terminating on the two LTP degen resistors, dropping around 20 volts, and support the cascode bases from here. Bump up the existing LTP CCS by 2mA to maintain the same operating stage current. This will ensure constant voltage across the LTP devices, reducing Early effect, but serve to reduce the Vbe they see.
Of course, Alex is right that cfp LTP would be the next logical step.... with changes in Ic, the Vbe changes cause a gross non-linearity in the LTP. With high fb factor, this is small, but it can be almost eliminated with a cfp.
Nice circuit, OS, looks a lot like the Hitachi AN of 1972 for their new-fangled mosfets!
Cheers,
Hugh
How about doing a constant power cascode?
That is, string a 2mA CCS from the positive rail to a resistor of 10K terminating on the two LTP degen resistors, dropping around 20 volts, and support the cascode bases from here. Bump up the existing LTP CCS by 2mA to maintain the same operating stage current. This will ensure constant voltage across the LTP devices, reducing Early effect, but serve to reduce the Vbe they see.
Of course, Alex is right that cfp LTP would be the next logical step.... with changes in Ic, the Vbe changes cause a gross non-linearity in the LTP. With high fb factor, this is small, but it can be almost eliminated with a cfp.
Nice circuit, OS, looks a lot like the Hitachi AN of 1972 for their new-fangled mosfets!
Cheers,
Hugh
How you doing ,Hugh.
Nice circuit , but MJL beat me to it... A perfect match for my 70's style board..Mouser stuff (big pile of cheap KSA 1845/1381/3503's)
just left texas...
Yes, already in the conception process.. But
with a new twist. I have a grossly underpowered Peerless
12" sub and want to build a 6/8 OP triple EF (aka leach)
but with my frugal parts ($1.40 0282/0302's). Should
be a good design for sub 100hz use as it can drive very low
imp. loads. Also it will let me have a crack at this topology
rounding out my experiences... The frugalbeast will be
born.
I almost have finalized the protection circuit for all to discuss.
OS
Love the discussion thats been going on over the Frugalamp II design. This is pretty much the topology I'd simmed a while back and then never got around to doing anything with.
I might have to lay out a PCB I know MJL has done one, but it wouldn't fit with how I plan to mount the transistors to my heatsinks. The cascode in the LTP is something I've had from day one, too.
I'm tempted to try simulating a CFP LTP input but i doubt it's worth the hassle.
I might have to lay out a PCB
I know MJL has done one, but it wouldn't fit with how I plan to mount the transistors to my heatsinks. The cascode in the LTP is something I've had from day one, too.I'm tempted to try simulating a CFP LTP input but i doubt it's worth the hassle.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.