I don't have ECC99 in spice model but I put in a 5687. You need about +/-18V sinewave on input to clip. Just about medium-mu triode will work with enough plate dissipation and anode voltage as the bias is set by the CCS and so the plate voltage will always be the same no matter which device you pick.
You will improve the HF performance a lot with the new driver. However this will become problematic with the NFB as the dominate pole will be too high now you don't have a weedy driver. The solution will be to add it in as a series R and C with the junction 47K of the plate of the EF86 to ground. I would guess you will need 3K3 and 330p for a 10KHz break point.
You will improve the HF performance a lot with the new driver. However this will become problematic with the NFB as the dominate pole will be too high now you don't have a weedy driver. The solution will be to add it in as a series R and C with the junction 47K of the plate of the EF86 to ground. I would guess you will need 3K3 and 330p for a 10KHz break point.
Last edited:
Awesome. I don't have the required components at home at this time so it will have to wait for next week. I will buy the stuff (especially the transistor cause I have no idea what to go with since I don't have a stockpile of them) and then see the result.
I will probably paly arround with the RC component in the plate of the EF86 see where it works best and what is the break point of going into oscillation. I do have a 12BH7 so that's why I wanted to try it with that. Would be more desirable.
I will probably paly arround with the RC component in the plate of the EF86 see where it works best and what is the break point of going into oscillation. I do have a 12BH7 so that's why I wanted to try it with that. Would be more desirable.
The good thing is you can try out different valves in the circuit without change the bias components.
On NFB There's some good articles on this.
If you have a dual channel scope you can open up the NFB and do a bode plot. That way you take the guesswork out. Without sounding critical many home brew amps end up being marginally stable.
This is good read...
Amplifier Compensation.
On NFB There's some good articles on this.
If you have a dual channel scope you can open up the NFB and do a bode plot. That way you take the guesswork out. Without sounding critical many home brew amps end up being marginally stable.
This is good read...
Amplifier Compensation.
Oh absolutely I will go trought that post. I have a 4channel digital HP scope (out of which two channels are 100% not limited in any ways) so yeah definitely will to improve my prototype one so in the future if I get to make another amplifier or maybe for someone else then I make a guaranteed working and stable amplifier. Very much apprechiated
Interesting reading. I had looked at the site covering the bode plots before.
Shock of the day, I replaced the finder relay octal sockets with some belton items. There was a moderate improvement in HF performance out of the phase splitter. That said I'm still not sure it's as good as it should be. These tests are done open loop with feedback left out of circuit CH1 is the output of the one anode on V2, CH2 is the input signal
Before
After
I'm now using the elektor documentation to help make sure the DC conditions are right as they have listed the voltages on the claus byrith modified version which is very handy.
Valve Final Amp Part 1.pdf - Google Drive
Valve Final Amp Part 2_ printed circuit boards and construction.pdf - Google Drive
I've now changed all of the resistors to match the CB design so against the original mullard component numbers, R9 is now 100K instead of 270K and R17 is now 10K instead of 15K. The anode load on V1 is about 53K rather than 47K. (Closest ≥0.5W resistor I have) These are the voltages I measured during today's test with all of these values as per the elektor/CB design. (all bar the manual balance/bias)
V1 - EF86
A - 98.4V
K - 0.98V
V2 - 12AX7
A1 - 287V
A2 - 294V
G1 - 89V
G2 - 98.4V
K - 100V
V3 - KT77
A - 433V
SG - 431V
K -31.2V
V4 - KT77
A - 431.5V
SG - 429.6V
K - 31V (moving around between 30-32V)
C15 - 454V
C12 -440V
C5 - 406V
C4 -206.5V
Shock of the day, I replaced the finder relay octal sockets with some belton items. There was a moderate improvement in HF performance out of the phase splitter. That said I'm still not sure it's as good as it should be. These tests are done open loop with feedback left out of circuit CH1 is the output of the one anode on V2, CH2 is the input signal
Before
After
I'm now using the elektor documentation to help make sure the DC conditions are right as they have listed the voltages on the claus byrith modified version which is very handy.
Valve Final Amp Part 1.pdf - Google Drive
Valve Final Amp Part 2_ printed circuit boards and construction.pdf - Google Drive
I've now changed all of the resistors to match the CB design so against the original mullard component numbers, R9 is now 100K instead of 270K and R17 is now 10K instead of 15K. The anode load on V1 is about 53K rather than 47K. (Closest ≥0.5W resistor I have) These are the voltages I measured during today's test with all of these values as per the elektor/CB design. (all bar the manual balance/bias)
V1 - EF86
A - 98.4V
K - 0.98V
V2 - 12AX7
A1 - 287V
A2 - 294V
G1 - 89V
G2 - 98.4V
K - 100V
V3 - KT77
A - 433V
SG - 431V
K -31.2V
V4 - KT77
A - 431.5V
SG - 429.6V
K - 31V (moving around between 30-32V)
C15 - 454V
C12 -440V
C5 - 406V
C4 -206.5V
Last edited:
That is a 10KHz square wave but not measured on the speaker output. This is measured at the anode of the phase splitter vs the input. The only residual issue once the loop is closed currently is poor performance at 20KHz. The design should be able to manage at least 20W cleanly out to 30KHz. Mine starts to distort at around 12W. I need to double check what I've seen, but I believe it is primarily one one half of the waveform only.
Like this, a kink starts to form near the positive peak of the sine wave. It doesn't seem to happen on the negative half. I still need to do some more testing to determine if it happens in the phase splitter or the output stage. (this shot is input signal vs speaker output signal)
Oh yes I see. Its quite a common problem - my K4040 had a similar shape actually much worse. It won't make that much difference to the amp musically. It may be the LTP is running out of stream with the miller capacitance. If that's the case you could drop the LTP plate resistors to say 68k and go with a 12AT7 LTP. You will need to drop the cathode resistor and the gain will drop which will need the NFB adjusting. Quite a lot of change. Interesting - nice to see you have the equipment to test and lots of photos.
Last edited:
Just simulated this and get the same result. You are running out of drive when the voltage on the LTP plate gets near 400v. You need about +/- 40v to drive EL34 fully.
The biggest improvement is to replace the 82k with a CCS. Increase the current a bit to get 250v on the plates and that seems to do the trick.
The biggest improvement is to replace the 82k with a CCS. Increase the current a bit to get 250v on the plates and that seems to do the trick.
I have both a single ECC81 and ECC82 for potential future experimentation. A CCS is somewhat beyond my scope though as I lack both the knowledge or parts to achieve it currently.
I've put some variable components in the feedback path in order to do some tests, partly to actually see for myself what each component has an effect on.
A 4.7K pot in the feedback path with a 680pF cap across it and the step network in with a 50K pot and a 3.5-20pF trim-cap.
The 4.7K pot has a huge effect as it controls the overall quantity of feedback so the amount of gain varies quite significantly when making changes. Observing the change in gain and phase vs frequency was nice and seeing what happens when there is too much NFB. Shame I don't have a variable cap big enough for this position.
Adjustment to the 50K pot seemed to have absolutely zero effect on the step network. The trim-cap had a very small effect on ringing on square waves but it was tiny.
This was the effect with feedback on a 10KHz square input. (500mV input voltage IIRC) Notably there appears to be some balance issues as the negative half has a hint of ringing that is absent in the positive and the signal magnitude differs slightly. At least I now know that 90% of the problems that I'm chasing are in the V2 stage.
just get any ol 100V transistortwo resistors one zener one small cap bam you have a current source. You can calculate the current by zener voltage - 0,5V divide by emitter resistor and it ill be roughly close enough for what you need anyways 
.
it easier than modding the entire V2 stage to a differet tube. Much faster at any rate. I suggest you try it before completely changing out the V2 circuit
.it easier than modding the entire V2 stage to a differet tube. Much faster at any rate. I suggest you try it before completely changing out the V2 circuit
Given what you have built a CCS is well within reach. The Ltspice schematic has one in which has the correct values for you. You may not want a transistor in your design, but it seems the easiest fix for you getting enough drive level and symmetry at 20KHz. Another option is an EF86 as a CCS.
https://drtube.com/datasheets/ef86-sed1996.pdf
https://drtube.com/datasheets/ef86-sed1996.pdf
Last edited:
My opinion just go with the transistor CCS. It only makes things better and it's a bunch cheaper and you never have to ever replace a tube.
If you want to make a CCS out of a vacuum tube (and I am not sure how to design such a circuit at all) Mr.Carlson's Lab has a video of fixing two "six pac" amplifiers. They used a EL84 as current source in the circuit
If you want to make a CCS out of a vacuum tube (and I am not sure how to design such a circuit at all) Mr.Carlson's Lab has a video of fixing two "six pac" amplifiers. They used a EL84 as current source in the circuit
Looking back over this thread, the general recommendation is to get shot of 12AX7 and replace it with something superior.
These are the current 9 pin options mentioned:
12AT7 - More gain than AU7. Is it more linear and does it require a CCS?
12AU7 - Requires CCS to get it linear and a lot less gain.
12BH7 - Linear without CCS but needs twice the heater current. (Fine with me as my two noval tubes are on a 2.5A heater tap)
ECC99 - No idea, but in principal, I have sufficient spare heater current that I could run one of these.
As to the CCS, I assume that this could be built on some veroboard and "tagged in" between two turrets. I'm not one of those who are bothered by using solid state parts in a valve amp. A hybrid is fine by me, so long as it performs. I've had a quick look through what I have. The only zener I have floating around is 1N5339BG. I've got some 2N3055's but prefer not using T-03 devices due to their size. I might have some BD139's, but they look like their too low voltage. Probably easier to just buy something if necessary. How much heat would this need to dissipate?
These are the current 9 pin options mentioned:
12AT7 - More gain than AU7. Is it more linear and does it require a CCS?
12AU7 - Requires CCS to get it linear and a lot less gain.
12BH7 - Linear without CCS but needs twice the heater current. (Fine with me as my two noval tubes are on a 2.5A heater tap)
ECC99 - No idea, but in principal, I have sufficient spare heater current that I could run one of these.
As to the CCS, I assume that this could be built on some veroboard and "tagged in" between two turrets. I'm not one of those who are bothered by using solid state parts in a valve amp. A hybrid is fine by me, so long as it performs. I've had a quick look through what I have. The only zener I have floating around is 1N5339BG. I've got some 2N3055's but prefer not using T-03 devices due to their size. I might have some BD139's, but they look like their too low voltage. Probably easier to just buy something if necessary. How much heat would this need to dissipate?
Not a lot about 100v*2ma = 200mW. If you add a resistor in the collector we could drop some of the voltage with this. Do you have some small signal transistors (TO92 package). The zener is a big big W wise. Do you have any LED's. A BD139 would be fine we would need to drop a bit of voltage.
Last edited:
I've got plenty of bog standard low current antique red/green LED's from 20+ years ago. I also have a couple of super bright reds.
Just for reference regarding valve heater support. My transformer 6.3V heater windings are 4A and 2.5A. The 4A tap is being used for the EL34's only and the 2.5A tap is doing EF86 and ECC83.
Just for reference regarding valve heater support. My transformer 6.3V heater windings are 4A and 2.5A. The 4A tap is being used for the EL34's only and the 2.5A tap is doing EF86 and ECC83.
Yep you can build your CCS with an LED and BD139, feed about 1ma from the HT (say 470K) into the LED. This will generate about 1.5V depending on the colour you use. Add a cap across the LED. Connect to base of NPN. Emitter resistor on NPN sets current. So R = (1.5-.65)/Iset. Iset is about 2ma I think as you want 250v on plates of V2 and half the current flows through each. Add collector resistor to drop a bit of the voltage (say 40v) to bring into range of BD139. Say 40v/2ma = 22k. If you need schematic I can get one. You may not need the veroboard - may just build it around the BD139. There's only a small dissipation on the transistor you could use say a TO92 package.
Last edited: