I've read through JB and Merlin Blencowe's books and through many different post regarding using CCS in the LTP. I'm very excited about making this CCS but I think I'm doing something wrong?
First attachment is from here on Diyaudio from some other members using a 10m45s ccs. I made that one first and checked my voltage and current and they were too high around 130-140 volts on cathode and around 14.5ma almost 7ma on each plate to ground. But only about 10 - 15 volts dropped across the 56k plate resistors.
So I thought maybe the CCS I bought were not right. I'm living in China now and the sources are sometime questionable.
So I built attachment 2 using a BD139 and BC549C voltages as follows
BD139 c 132v b 3.37v e 2.82v
BC549c C 2.82v b 1.69 and e 1.09 still very high voltage on cathodes and only around 12volts dropped across the 56k resistors.
I added the jolida schematic and I removed the 15k resistor and that is were I placed the CCS. On the second schematic I tied the 270k resistor to the B= for the 12AT7. I am thinking I should be able to adjust the ma from around 4-8ma but no way and that I should be able to adjust the cathode voltage to around 100v. But this is not possible without the voltage drop across the plate resistors.
I've never worked with a CCS source before and I know it probably something stupid I'm doing or something I don't understand. Please shed some light on this poor soul
First attachment is from here on Diyaudio from some other members using a 10m45s ccs. I made that one first and checked my voltage and current and they were too high around 130-140 volts on cathode and around 14.5ma almost 7ma on each plate to ground. But only about 10 - 15 volts dropped across the 56k plate resistors.
So I thought maybe the CCS I bought were not right. I'm living in China now and the sources are sometime questionable.
So I built attachment 2 using a BD139 and BC549C voltages as follows
BD139 c 132v b 3.37v e 2.82v
BC549c C 2.82v b 1.69 and e 1.09 still very high voltage on cathodes and only around 12volts dropped across the 56k resistors.
I added the jolida schematic and I removed the 15k resistor and that is were I placed the CCS. On the second schematic I tied the 270k resistor to the B= for the 12AT7. I am thinking I should be able to adjust the ma from around 4-8ma but no way and that I should be able to adjust the cathode voltage to around 100v. But this is not possible without the voltage drop across the plate resistors.
I've never worked with a CCS source before and I know it probably something stupid I'm doing or something I don't understand. Please shed some light on this poor soul
I don't know which attachment is which -- could you be more descriptive than first, second, etc.? I think the forum software reordered them. Anyhow, something is not right with your measurements. 7mA does not drop 10-15V across 56K.
Something to remember about CCSes is that they are really current limiters. They won't force current where they can't.
With the 10M45, it is really a depletion mode mosfet. There is a voltage from the gate to the source, and the fet attempts to maintain that voltage. So, if it is -4.5V, then a 100R resistor will cause it to operate as a 45mA CCS. The actual voltage can vary quite a bit from chip to chip (and is not constant for all currents, but it is close enough for this use), so you need to either measure Vgs, or use a trimmer.
Something to remember about CCSes is that they are really current limiters. They won't force current where they can't.
With the 10M45, it is really a depletion mode mosfet. There is a voltage from the gate to the source, and the fet attempts to maintain that voltage. So, if it is -4.5V, then a 100R resistor will cause it to operate as a 45mA CCS. The actual voltage can vary quite a bit from chip to chip (and is not constant for all currents, but it is close enough for this use), so you need to either measure Vgs, or use a trimmer.
A few things to consider:
The input-grid of the LTP is hanged on the anode of the pre-amp and pretty fixed.
Juggling around with a CCS in the LTP-cathode is not going to produce the results you expect.
The input-triode of the LPT needs to be operating with a sensible bias, something like -2V (Cathode 2V above the grid, depending on tube-types, desired sound-quality warm or cold etc)
In the original Jolida circuit if you have around 100V on grid-1 the cathode should also land around here (+2V) and that gives a LTP-current through the original 15K tail-resistor of 102/15K = 6,8mA That's what the circuit was designed for at quiescent and I would dimension the CCS for that. It would leave the bias-setup unchanged.
Read the attached Morgan Jones pages on CCS for dimensioning and take a good look at the BD139 Datasheet attached. You are NOT going to get away with using that as the Vceo is a mere 80V and you are measuring 100V across C-E already. Use a MJE340 and use a high-gain small-signal transistor for setting the current. I used a MPS-A06 with success, see the attached schematics for using a CCS in the tail of a Cathode-Follower used to drive output-tubes into AB2-operation. High gain in this transistor ends up in high impedance if the CCS, see the Morgan Jones attachment on calculating Rout.
The power-transistor is going to dissipate almost 1W and that will mean putting it on a small heat-sink.
rgds,
/tri-comp
The input-grid of the LTP is hanged on the anode of the pre-amp and pretty fixed.
Juggling around with a CCS in the LTP-cathode is not going to produce the results you expect.
The input-triode of the LPT needs to be operating with a sensible bias, something like -2V (Cathode 2V above the grid, depending on tube-types, desired sound-quality warm or cold etc)
In the original Jolida circuit if you have around 100V on grid-1 the cathode should also land around here (+2V) and that gives a LTP-current through the original 15K tail-resistor of 102/15K = 6,8mA That's what the circuit was designed for at quiescent and I would dimension the CCS for that. It would leave the bias-setup unchanged.
Read the attached Morgan Jones pages on CCS for dimensioning and take a good look at the BD139 Datasheet attached. You are NOT going to get away with using that as the Vceo is a mere 80V and you are measuring 100V across C-E already. Use a MJE340 and use a high-gain small-signal transistor for setting the current. I used a MPS-A06 with success, see the attached schematics for using a CCS in the tail of a Cathode-Follower used to drive output-tubes into AB2-operation. High gain in this transistor ends up in high impedance if the CCS, see the Morgan Jones attachment on calculating Rout.
The power-transistor is going to dissipate almost 1W and that will mean putting it on a small heat-sink.
rgds,
/tri-comp
So neither of these CCS sources are suitable?
I uploaded the second CCS source using the BD139 BC549C. I will look through MJ book again I do have it. I tried the other two CCS as I thought they would work. Why is it there is so little voltage drop across the plate resistors?
On the original 15k tail resistor I had 102 on the grid of the at7 and 106 on the cathode of the tubes and more than 100v dropped across a 24k resistor? Sorry if my questions may elementary, I am truly trying to understand this.
Thanks Tricomp I will look at sources the transistors you mentioned. I was just curious why both of these CCS acted in the same way and I couldn't get the voltage drop across the plate resistors.
I was trying to get away with using the positive supply. On another thread SY recommended tying the base to the B+ and using the 270 k resistors. Do you think the 10M45s were also bad? They seem to work up to 450v. I was a little leery when I bought them here in China but if I didn't get them here in China I would have had to order from back in the U.S. Or would they not work for this application? It acted almost the same as the transistors in the CCS. I do see that 100v is not suitable for the BD139
I uploaded the second CCS source using the BD139 BC549C. I will look through MJ book again I do have it. I tried the other two CCS as I thought they would work. Why is it there is so little voltage drop across the plate resistors?
On the original 15k tail resistor I had 102 on the grid of the at7 and 106 on the cathode of the tubes and more than 100v dropped across a 24k resistor? Sorry if my questions may elementary, I am truly trying to understand this.
Thanks Tricomp I will look at sources the transistors you mentioned. I was just curious why both of these CCS acted in the same way and I couldn't get the voltage drop across the plate resistors.
I was trying to get away with using the positive supply. On another thread SY recommended tying the base to the B+ and using the 270 k resistors. Do you think the 10M45s were also bad? They seem to work up to 450v. I was a little leery when I bought them here in China but if I didn't get them here in China I would have had to order from back in the U.S. Or would they not work for this application? It acted almost the same as the transistors in the CCS. I do see that 100v is not suitable for the BD139
For any reasonable current, the adjustment pot value is a bit high. Let's say (for example) that you want 2mA per 12AT7 section. That's 4mA tail current. If your red LED drops 1.7V (check that!), we subtract off the 0.7V Vbe drop in the transistor, then calculate the emitter resistor as R = 1V/4mA = 250R. If you want it adjustable, a 500R trimmer would be more appropriate. For higher currents, that resistor value will obviously be smaller.
OK, I looked at your schematic and SY is of course right. But why not just calculate the right fixed value and put it in?
Then another thought hit me. You already have a -80V supply for the output-bias. You could stabilize that with a TL783 (See attachment) for something like -60V which is fine for the output-bias circuitry (resistor-values here needs to be re-calculated for the 20V lower supply as well as the value of R24) Now use the -60V as input for the CCS, do away with the LED and use a regular silicon-diode for compensating the small signal transistor and a zener-diode like Morgan Jones (and I, being inspired
) did it. Calculate the right value of the emitter resistor for 6,8mA CCS-current and you'll be set.
I'll attach the power-supply for the above 2C34-PP amp so you get the formula for calculating output-voltage of the TL783
rgds,
/tri-comp
Then another thought hit me. You already have a -80V supply for the output-bias. You could stabilize that with a TL783 (See attachment) for something like -60V which is fine for the output-bias circuitry (resistor-values here needs to be re-calculated for the 20V lower supply as well as the value of R24) Now use the -60V as input for the CCS, do away with the LED and use a regular silicon-diode for compensating the small signal transistor and a zener-diode like Morgan Jones (and I, being inspired
) did it. Calculate the right value of the emitter resistor for 6,8mA CCS-current and you'll be set. I'll attach the power-supply for the above 2C34-PP amp so you get the formula for calculating output-voltage of the TL783
rgds,
/tri-comp
Thanks SY , Ok now I'm getting somewhere. I changed to a 500v trimmer and I can adjust the voltage on the cathodes into the 100v range. I now drop 270 volts across the plate resistors so 4.82ma.
That seems a bit high? Should I move to a 1k trimmer? Also I think I will need to lower the plate resistors back to where they first were at the 24K range? I appreciate your patience. I think it's closer now.
That seems a bit high? Should I move to a 1k trimmer? Also I think I will need to lower the plate resistors back to where they first were at the 24K range? I appreciate your patience. I think it's closer now.
Thanks tri-comp,
I will give that a shot as MJ says that it's a better CCS. This is my first go at a CCS and I wanted to give it a try and see how it works and sounds. As of now I don't have a TL783 or the MJE340. I will source them in Guangzhou when I go there for work and give them a try. I've been doing a lot of reading but some of this stuff is still hard for me to take in.
If I can get this set up with my current parts on hand first I will be very happy.
I will give that a shot as MJ says that it's a better CCS. This is my first go at a CCS and I wanted to give it a try and see how it works and sounds. As of now I don't have a TL783 or the MJE340. I will source them in Guangzhou when I go there for work and give them a try. I've been doing a lot of reading but some of this stuff is still hard for me to take in.
If I can get this set up with my current parts on hand first I will be very happy.
After replacing the 500r trimmer I can adjust with no problem to the 6.8ma which was in the original design with the 15k resistor. Only one last question.
On one side of the tube "input side" I have 357 on the plate 117.5 on the grid and 121.3 on the cathode. But on the other part of the tube it 362 v on the plate 107 volt on the grid and 121.3 on the cathode.
My question is I thought using the CCS would balance the two side out? What's with the 10 volt difference on the grids? Isn't that a problem? The 1meg resistor is 1meg. Should I be looking elsewhere? Thanks for your help.
On one side of the tube "input side" I have 357 on the plate 117.5 on the grid and 121.3 on the cathode. But on the other part of the tube it 362 v on the plate 107 volt on the grid and 121.3 on the cathode.
My question is I thought using the CCS would balance the two side out? What's with the 10 volt difference on the grids? Isn't that a problem? The 1meg resistor is 1meg. Should I be looking elsewhere? Thanks for your help.
Here's one thing to watch for in your tradeoffs between plate resistors and current- the signal swing. You want at least 150V between plate and grid of the 12AT7, and you want to drop at least 150V across the plate resistors. That means larger resistors = lower current or smaller resistors = higher current.
Thanks for your patience SY. Putting the 56k resistors back in I have adjusted the bias so that I have 167 volt drop on one side along with 163 volt plate to grid difference.
The other side is 162 volt drop and 186 volt plate to grid difference. It falls into what you have said. Isn't the almost 20 volt difference on the grids important or is that not an issue? It gives me around 3 ma per side. Is everything kosher?
The other side is 162 volt drop and 186 volt plate to grid difference. It falls into what you have said. Isn't the almost 20 volt difference on the grids important or is that not an issue? It gives me around 3 ma per side. Is everything kosher?
You can't measure the grid-voltage on the second tube in the LTP 100% correctly.
The grid-circuit impedance is far too high and your voltmeter will be loading the actual value down. Even with a 10Mohm/V hi-impedance voltmeter you will still be adding a resistor in parallel with the present 1Mohm grid-resistor and read voltage out of what's expected. And that's with a 10Mohm/V meter, most digital-meters have a 1Mohm input-impedance making things even worse.
Never take and trust grid-1 readings unless the grid is DC-driven from a low-impedance circuit like the input triode in the LTP or like the g1 of the 2C34 output triode in the 2C34-PP amp above. Here the g1 circuit is a low-impedance cathode-follower and readings will be valid.
3mA per side totals 6mA in the LTP which is very close to the estimated original 6,8mA. I wouldn't worry about the difference.
I WOULD worry about the output-impedance of the CCS (see MJ on impedance and how to calculate)
rgds,
/tri-comp
The grid-circuit impedance is far too high and your voltmeter will be loading the actual value down. Even with a 10Mohm/V hi-impedance voltmeter you will still be adding a resistor in parallel with the present 1Mohm grid-resistor and read voltage out of what's expected. And that's with a 10Mohm/V meter, most digital-meters have a 1Mohm input-impedance making things even worse.
Never take and trust grid-1 readings unless the grid is DC-driven from a low-impedance circuit like the input triode in the LTP or like the g1 of the 2C34 output triode in the 2C34-PP amp above. Here the g1 circuit is a low-impedance cathode-follower and readings will be valid.
3mA per side totals 6mA in the LTP which is very close to the estimated original 6,8mA. I wouldn't worry about the difference.
I WOULD worry about the output-impedance of the CCS (see MJ on impedance and how to calculate)
rgds,
/tri-comp
Last edited:
What tricomp said. The drops you're reporting look excellent. I think you're doing great.
One other thing- the impedance of your CCS is likely to be pretty high (that's good). What makes life even sweeter is that an LTP CCS doesn't have to be as good as a plate one because the impedance at the cathodes is pretty low in comparison.
One other thing- the impedance of your CCS is likely to be pretty high (that's good). What makes life even sweeter is that an LTP CCS doesn't have to be as good as a plate one because the impedance at the cathodes is pretty low in comparison.
Thank you guys very much. I've been up all night here with this thing but I'm really happy I'm learning something.
Tricomp your point about the BD139 is taken. I will look to source the MJE340. I will need to do a lot more reading to take in MJ out-put impedance math. I do note that he said the output impedance is greatly increased using the d design. "the one you recommended to me in the first place. I will finish the other channel and give them a listen tomorrow "uh" later today. Haven't slept yet.
Tricomp your point about the BD139 is taken. I will look to source the MJE340. I will need to do a lot more reading to take in MJ out-put impedance math. I do note that he said the output impedance is greatly increased using the d design. "the one you recommended to me in the first place. I will finish the other channel and give them a listen tomorrow "uh" later today. Haven't slept yet.
Spoken like a real tube-geek
Don't always trust what you see on schematics, people make errors. Even MJ on his proposed CCS-designs. Exactly in the 'd'-circuit the wrote the wrong voltage-reading. The MJE340-Emitter/BC549-Collector junction should read 134V, not 124. Just goes to show errors creep in where you least expect them to.
I'll help you calculate the CCS if you want but doing it yourself and posting the result here will be more educational
rgds,
/tri-comp
Don't always trust what you see on schematics, people make errors. Even MJ on his proposed CCS-designs. Exactly in the 'd'-circuit the wrote the wrong voltage-reading. The MJE340-Emitter/BC549-Collector junction should read 134V, not 124. Just goes to show errors creep in where you least expect them to.
I'll help you calculate the CCS if you want but doing it yourself and posting the result here will be more educational
rgds,
/tri-comp
Well after a few hours asleep and some time with the wife, "if not she would fuss".
I have to say wow I am really happy with the results. Wish I would have done this a long time ago. I no longer feel the image is slightly off, it's really is solid and balanced. I was expecting the information retrieval to be improved because of the loss of distortion. But what I didn't expect was the feeling of naturalness. Sorry can't think of a better word.
I set both sides within the parameters in which SY provided. Here is where it sits.
Using 56k resistors "matched" I have about 155-162 dropped across the plate resistors.
152 to 155 volt differences between plate and grid and about 2.92-2.95 ma current each side.
I am still not comfortable with the BD139 voltage too high. I will try to source the parts very soon. Leaving for Guangzhou tomorrow and then will be busy for a few days.
tricomp I will read some more about the output impedance of the CCS and how to figure it out mathematically. I'm still learning and need to understand how it is calculated and how it effects the next stage and how to optimize it. I will do some reading.
By the way thanks you guys I learned something and I'm having fun.
I have to say wow I am really happy with the results. Wish I would have done this a long time ago. I no longer feel the image is slightly off, it's really is solid and balanced. I was expecting the information retrieval to be improved because of the loss of distortion. But what I didn't expect was the feeling of naturalness. Sorry can't think of a better word.
I set both sides within the parameters in which SY provided. Here is where it sits.
Using 56k resistors "matched" I have about 155-162 dropped across the plate resistors.
152 to 155 volt differences between plate and grid and about 2.92-2.95 ma current each side.
I am still not comfortable with the BD139 voltage too high. I will try to source the parts very soon. Leaving for Guangzhou tomorrow and then will be busy for a few days.
tricomp I will read some more about the output impedance of the CCS and how to figure it out mathematically. I'm still learning and need to understand how it is calculated and how it effects the next stage and how to optimize it. I will do some reading.
By the way thanks you guys I learned something and I'm having fun.
The output impedance of the CCS is calculated by hoe + (hfe1hfe2)Re, and is likely to be well over 1M. hoe is rarely on the datasheets, but can be approximated by looking at the slopes of the collector current versus collector voltage curves. Generally, the second term dominates if the transistors have a high hfe.
As I mentioned before, in a cathode circuit, the CCS source impedance isn't terribly critical. If memory serves, Morgan Jones's book has a table of hoe for some common transistors, so you can come up with a pretty decent approximation.
As I mentioned before, in a cathode circuit, the CCS source impedance isn't terribly critical. If memory serves, Morgan Jones's book has a table of hoe for some common transistors, so you can come up with a pretty decent approximation.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.