Phil,
R14 is 470 Ohms, not 470 KOhms. A Mills MRA-5 is appropriate here too.
C1 is outside the global NFB loop. Therefore, it has an impact on the amp's voice. A paper in oil (PIO) part gets my nod. Soviet surplus is the best deal in PIO capacitors. Regardless of the cap. construction settled on, a well matched pair of parts (for the 2 channels) is in order.
Low WVDC 'lytics are ubiquitous. Think of all the digital stuff that runs on 5 V. Look at this Nichicon data sheet, for a suitable part. BTW, 6 WVDC is minimum. Using a part that's rated for more Volts is perfectly acceptable.
We use the interlocked circles to represent the conglomeration of parts that comprise the CCS. The box for the 10M45S (by itself) is fine. Only 2 passive parts are needed, along with the 10M45S. A 1 KOhm trim pot. is used to set the current. A 100 Ohm 1% tolerance metal film part is used as a test point. The trim pot. is wired up as a simple variable resistor and connects to the K terminal and the 100 Ohm part. The G terminal connects to the junction of the pot. and test resistor. Start with the pot. set for max. resistance and back off until the drop measured across the 100 Ohm resistor is 0.3 V.
R14 is 470 Ohms, not 470 KOhms. A Mills MRA-5 is appropriate here too.
C1 is outside the global NFB loop. Therefore, it has an impact on the amp's voice. A paper in oil (PIO) part gets my nod. Soviet surplus is the best deal in PIO capacitors. Regardless of the cap. construction settled on, a well matched pair of parts (for the 2 channels) is in order.
Low WVDC 'lytics are ubiquitous. Think of all the digital stuff that runs on 5 V. Look at this Nichicon data sheet, for a suitable part. BTW, 6 WVDC is minimum. Using a part that's rated for more Volts is perfectly acceptable.
We use the interlocked circles to represent the conglomeration of parts that comprise the CCS. The box for the 10M45S (by itself) is fine. Only 2 passive parts are needed, along with the 10M45S. A 1 KOhm trim pot. is used to set the current. A 100 Ohm 1% tolerance metal film part is used as a test point. The trim pot. is wired up as a simple variable resistor and connects to the K terminal and the 100 Ohm part. The G terminal connects to the junction of the pot. and test resistor. Start with the pot. set for max. resistance and back off until the drop measured across the 100 Ohm resistor is 0.3 V.
Sorry, I'm swamped with yet another task (paying the price for goofing off last night) so I haven't checked the values of your cathode resistors. But the grid stoppers for the output tubes are probably too big for triode mode (lots of Miller effect). 1-2k should be fine there.
You should probably consider bypassing the 470R cathode resistor. And to make the input simpler, you could either direct couple (do you expect any offset voltages from your sources?) or at minimum use standard values for the input cap and grid leak resistor like 100n and 1M. Eli is correct about the fundamental frequencies of bass, but you really want your f3 a decade lower than the lowest frequency of interest to reduce phase shift.
I do not like paper caps at all (they can be leaky and drify, very humidity sensitive), but others do. If it were my amp, I'd use a good polypropylene and foil like Wima FKP or some of those cool Russian surplus Teflon caps.
You should probably consider bypassing the 470R cathode resistor. And to make the input simpler, you could either direct couple (do you expect any offset voltages from your sources?) or at minimum use standard values for the input cap and grid leak resistor like 100n and 1M. Eli is correct about the fundamental frequencies of bass, but you really want your f3 a decade lower than the lowest frequency of interest to reduce phase shift.
I do not like paper caps at all (they can be leaky and drify, very humidity sensitive), but others do. If it were my amp, I'd use a good polypropylene and foil like Wima FKP or some of those cool Russian surplus Teflon caps.
Phil, there you have it. R3, R4, R15, and R18 are all 1 KOhm Carbon composition. 733 Ω is the calculated value for R . 732 Ω is the closest "stock" value. 715 Ω is the next lower "stock" value and should be safe to use too. If you bypass the 470 Ω EL42 shared cathode resistor, use a 50 WVDC/150 μF. part.
SY, rolling the deep bass off slightly below 30 Hz. is INTENTIONAL. The Edcor O/P trafos Phil is ordering will require a substantial LF error correction signal to extend their bass response. An overly large correction signal will saturate the O/P trafo core.
IMO, the selection made is the best compromise. FWIW, we did the same thing in "El Cheapo" and achieved a satisfactory result.
I'd still rather see an LED there. I hate electrolytics in series with the input signals, though admittedly it's common practice. I haven't pulled out the datasheets and done a loadline analysis, so I'll assume that Eli (a very competent guy) did it right. Worst case, the plate voltage you get is somewhat off and you have to change two resistors. No big deal.
Also, you will probably have to add some compensation circuitry to get stability, but that's something you'll have to determine empirically. One school of thought (mine) is to not worry too much about squeezing out the last bits of HF badwidth, so the amp is compensated without a feedback capacitor, just by using a series RC step network across the plate resistor of the first stage. Starting values would be something like 10% of the plate resistor value and a capacitance to give a time constant corresponding to 10kHz. You'll want to use a scope and square wave generator with a resistive and a reactive dummy load; this is well covered in MJ's "Building Valve Amplifiers."
Others prefer using a capacitor across the feedback resistor. Crowhurst argues against that in several of his books and his arguments look sound (no pun intended). In "BVA," MJ recommends a combination of the two methods, and since he's smarter than me, he's likely to be right.
Also, you will probably have to add some compensation circuitry to get stability, but that's something you'll have to determine empirically. One school of thought (mine) is to not worry too much about squeezing out the last bits of HF badwidth, so the amp is compensated without a feedback capacitor, just by using a series RC step network across the plate resistor of the first stage. Starting values would be something like 10% of the plate resistor value and a capacitance to give a time constant corresponding to 10kHz. You'll want to use a scope and square wave generator with a resistive and a reactive dummy load; this is well covered in MJ's "Building Valve Amplifiers."
Others prefer using a capacitor across the feedback resistor. Crowhurst argues against that in several of his books and his arguments look sound (no pun intended). In "BVA," MJ recommends a combination of the two methods, and since he's smarter than me, he's likely to be right.
"Perfect" phase compensation is very tricky business and requires a sophisticated test bench. I favor a brute force approach to the problem that requires a builder own only a good multi-meter.
I like to "peak" the voltage amplifying block's HF behavior, by the use of inductive wire wound load resistors. Unfortunately, the CCS loading necessary in this design rules that out. Please notice that the 1800 pF. cap. from "hot" speaker lead to ground shorts the NFB loop out beginning at approx. 80 KHz. Instead of trying to squeeze too much performance out, the natural roll of the O/P trafo is allowed to set in. The measures I advocate hold the HF error correction signal to (hopefully) a reasonable level and avoid slew limiting. The high gm of the 6922 small signal tube in this design is definitely an asset.
Yeah, I forgot about the CCS, but I still think a step network across it will do a fine job of stabilizing things. I like to adjust for stability on a resistive load, then hang a 2u cap across the load resistor and see if the ringing gets to be too much. Typically, with a moderate feedback design like this, I'd aim for an open loop -3dB point of 5-10 times lower than the transformer's HF resonance; it's a variation of dominant pole compensation, and one is much more likely to end up with an amp that will remain stable with a real-world speaker load. The trade-off, as I mentioned earlier, is HF bandwidth; the evidence that high bandwidth is a significant factor in audibility is sketchy at best, but without question, and amp that is marginally stable will be annoying to listen to.
Although it is possible to roughly adjust stability without a scope, I wouldn't recommend it. Why work blind?
Although it is possible to roughly adjust stability without a scope, I wouldn't recommend it. Why work blind?
Hi all, not had much to say for a while because I've been busy collecting parts for this project... I'm having trouble finding the 10M45S's in the UK though, does anyone have a couple spare they would be willing to sell? please?
BTW thanks for the advice on possible stability issues, my copy of 'building valve amplifiers' is in the post, I will be reading up on these issues and although I hope it won't be necessary, I do have access to a 'scope at work if needed
Phil.
BTW thanks for the advice on possible stability issues, my copy of 'building valve amplifiers' is in the post, I will be reading up on these issues and although I hope it won't be necessary, I do have access to a 'scope at work if needed
Phil.
The 10M45S is the simplest CCS load, but plenty of decent designs are available. IIRC, forum member "Pinkmouse" is a UK resident. Perhaps he or another UK resident member will part with a pair of CCS PCBs that were group ordered, some time ago. Those boards should be populated with "P" type BJTs, for this job.
How are you getting on with this Phil? I think you may be able to use any old voltage regulator for the CCS. A 7812 with 4.7k set resistor would be worth a shot.phil_2 said:Hi all, not had much to say for a while because I've been busy collecting parts for this project... I'm having trouble finding the 10M45S's in the UK though, does anyone have a couple spare they would be willing to sell? please?
BTW thanks for the advice on possible stability issues, my copy of 'building valve amplifiers' is in the post, I will be reading up on these issues and although I hope it won't be necessary, I do have access to a 'scope at work if needed
Phil.
Aslo, youmay need to add pole slugging caps in parallel with the power valve's grid leak resistors to maximise stability. I'm thinking 47pF will be in the right area.
Thanks for the heads up guys 
I've actually bought some CCS kits using the 10M45Ss off ebay now to keep the project moving. I'm also waiting for a whole bunch of other parts to arrive before I can start building, I'm keen to get started I will try the caps as you suggest Merlinb and I'll keep this thread updated as things progress.
I've actually bought some CCS kits using the 10M45Ss off ebay now to keep the project moving. I'm also waiting for a whole bunch of other parts to arrive before I can start building, I'm keen to get started I will try the caps as you suggest Merlinb and I'll keep this thread updated as things progress.Hi Guys, sorry its been a while, don't think that I've given up or anything, not for a minute! My output transformers have finally arrived and so have the other parts. In the mean time I have read 'Building Valve Amplifiers' which I purchased first, and I'm well into 'Valve amplifiers' which I bought second as I realised while reading 'BVA' that I needed it! Great books both of them. I've also been reading heaps of threads on this forum.
I have planned out my chassis and should be cutting the sheet aluminium on the weekend, all being well.
One question while I'm planning out switch locations etc, Is it worth including a triode/UL switch? will the amp work in UL mode? or is it likely to be unstable? I notice the output transformers have 40% taps...
Thanks again for all your help so far.
Phil.
I have planned out my chassis and should be cutting the sheet aluminium on the weekend, all being well.
One question while I'm planning out switch locations etc, Is it worth including a triode/UL switch? will the amp work in UL mode? or is it likely to be unstable? I notice the output transformers have 40% taps...
Thanks again for all your help so far.
Phil.
Hello again, I just wanted to post a few pictures just to show that progress is being made Obviously it now needs wiring up and also I'm yet to make a wooden frame for it yet. I'm pleased with how its coming together, hope you like it too. more pics to follow as things progress, which shouldn't be very long.
Phil.
Obviously it now needs wiring up and also I'm yet to make a wooden frame for it yet. I'm pleased with how its coming together, hope you like it too. more pics to follow as things progress, which shouldn't be very long.
Phil.
Hi guys.
I need some help please...Eli!? I have wired one channel and checked my wiring before powering up. I can hear music VERY faintly on the output, other symptoms are that R9, the 18.2k metal film resistor gets very hot and smells (it hasn't burnt out yet tho as I haven't run it for long) and the CCS, when measureing across the 100 Ohm resistor I get 2.2V = 2.2mA no matter where the pot is set? any ideas what the cause of these problems is? are there any other measurements I can take to help identify the problem?
thanks. phil.
I need some help please...Eli!? I have wired one channel and checked my wiring before powering up. I can hear music VERY faintly on the output, other symptoms are that R9, the 18.2k metal film resistor gets very hot and smells (it hasn't burnt out yet tho as I haven't run it for long) and the CCS, when measureing across the 100 Ohm resistor I get 2.2V = 2.2mA no matter where the pot is set? any ideas what the cause of these problems is? are there any other measurements I can take to help identify the problem?
thanks. phil.
I've heard mention of setting/testing CCS's independantly (I think it was Tubelab?) by connecting them across a 9v battery, is it safe to do this with the 10M45S with no other load than the 100 Ohm resistor?
Also, regarding the cathode resistor R9 mentioned above, could it just be that the resistor isn't up to the power rating required? I've used a 0.75W resistor here.
Thanks. Phil.
Also, regarding the cathode resistor R9 mentioned above, could it just be that the resistor isn't up to the power rating required? I've used a 0.75W resistor here.
Thanks. Phil.
I've got music!!!
I had the CCS wired wrong... in my defence the kit bought off ebay had two holes marked 'A' its quite confusing...
The cathode resistor for the phase splitter now runs cool aswell, so it looks like both problems are solved.
I can't really vouch for soung quality yet, I've only used a small test speaker so far. I'll wire the other channel and let you know.
I had the CCS wired wrong... in my defence the kit bought off ebay had two holes marked 'A' its quite confusing...
The cathode resistor for the phase splitter now runs cool aswell, so it looks like both problems are solved.
I can't really vouch for soung quality yet, I've only used a small test speaker so far. I'll wire the other channel and let you know.
Just a bit of an update, and feedback for those who helped me with this.
Both channels are now built and I've spent a bit of time listening to the amplifier. Baring in mind that none of my equipment could be called Hi-end and my speakers are actually small bookshelf speakers with 85db efficiency, with a bad low frequency response... the sound is good, very nice I was worried about the concerns expressed about low end frequency response and potential stability issues. firstly the amp is stable and produces sufficient power for my small (very small) listening room.
Secondly, bypassing the 12nf input caps shows little noticeable improvement in bass response (only just about noticeable in deep dub style reggae bass lines) probably because of my speakers inability to play the really deep stuff anyway. Bypassing them did however noticeably increase distortion at higher volume levels so I've left them in. Limiting the low end bandwidth was a worthwhile compromise in this case.
So to conclude I would like to do something I've been waiting to do, but couldn’t until I had a working amplifier; and say a BIG thank you to those who helped me out here, especially Eli. Its been a really fun project
When I finally get my cheap circular saw to behave and cut the right angles, or give up and pay someone to cut them for me... I will be able to finish off the wooden base and post a pic of the finished amplifier.
Phil.
Both channels are now built and I've spent a bit of time listening to the amplifier. Baring in mind that none of my equipment could be called Hi-end and my speakers are actually small bookshelf speakers with 85db efficiency, with a bad low frequency response... the sound is good, very nice
I was worried about the concerns expressed about low end frequency response and potential stability issues. firstly the amp is stable and produces sufficient power for my small (very small) listening room.Secondly, bypassing the 12nf input caps shows little noticeable improvement in bass response (only just about noticeable in deep dub style reggae bass lines) probably because of my speakers inability to play the really deep stuff anyway. Bypassing them did however noticeably increase distortion at higher volume levels so I've left them in. Limiting the low end bandwidth was a worthwhile compromise in this case.
So to conclude I would like to do something I've been waiting to do, but couldn’t until I had a working amplifier; and say a BIG thank you to those who helped me out here, especially Eli. Its been a really fun project
When I finally get my cheap circular saw to behave and cut the right angles, or give up and pay someone to cut them for me... I will be able to finish off the wooden base and post a pic of the finished amplifier.
Phil.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.