refference said:
Well, I build another circuit (not the one i'm modifying with your help) and it has no humm, no noise,.... "dead silent"
I hope that will happen the same with my new circuit.
thanks!!!
Hi zafira,
I just noticed that you have no grid resistor on the first stage! The input cap should go before the volume pot, not after. It's rather high resistance for a volume control but you can get away with it because the input capacitance of a pentode is very low. Maybe shunt the slider to ground with a 470k resistor, though, so the pot won't cause a scratchy noise in the speaker if its track has imperfections (I don't trust pots!)
I just spent some time trying to model your circuit with LTSpice , but to no avail. It seems the model I'm trying to use for KT88 is no good, it just seizes up! LTSpice can be a dog sometimes
. It works OK, though, with EL34s (and different voltages, of course). It shows that there could be excessive open-loop gain and it might work better without the cathode bypass cap. on stage 1, otherwise there could be LF instability.
The splitter operating points look OK with a CCS giving 6mA in the tail and 450v for B+, although 47k might be better than 56k for the plate loads.
I didn't use a cascode for the CCS in the model, I used a ring of two, which I find easier. If you do use a cascode, you're going to need to connect the base of the top transistor to something. It's probably OK having a 1k resistor (R9) to ground if you use a negative supply, as you have shown. However, without a negative supply, R9 must have a much higher value, to go between the base and B+. For safety's sake, a zener circuit would be a good idea but it seems a messy solution. So, your negative supply might be the best solution in the end, if you want to use cacoded transistors for a CCS!
HF instability could also rear its ugly head but that depends on your OP transformer more than anything else. You may have to experiment with values for Cf in approximtely the 150pF to 680pF range across Rf to get it right. Rf should be about 4.7k, I think, giving a closed loop gain of about 45. You can make it higher if you want more gain.
Your B+ at 520v seems a bit high for KT88s in AB1 UL. I would have thought 460v was more like it. The voltage for the splitter B+ at 520v seems too high as well. In any case, the splitter needs to be decoupled from the main B+ and a choke would do that nicely if you're going to drop B+ anyway; otherwise, use a 10k resistor. There has to be a decoupling cap. in there as well, whether you're using a choke or a dropping resistor: say 22uF to ground (after the dropping resistor or choke, not before it as you had last time).
The bias network on your OP tubes looks better now, although the range of adjustment you've allowed is rather a lot, from -80v to -40v (you need about -60). You can reduce the adjustment range by either increasing the resistor between the bias pot and ground (R17) or reducing the pot resistance, or both.
Don't forget to set the bias pots to the -80v end to start with, then adjust them carefully to give the required quiescent current. ~50mA seems right to start with but you need to adjust it so it sounds right to you.
I just noticed that you have no grid resistor on the first stage! The input cap should go before the volume pot, not after. It's rather high resistance for a volume control but you can get away with it because the input capacitance of a pentode is very low. Maybe shunt the slider to ground with a 470k resistor, though, so the pot won't cause a scratchy noise in the speaker if its track has imperfections (I don't trust pots!)
I just spent some time trying to model your circuit with LTSpice , but to no avail. It seems the model I'm trying to use for KT88 is no good, it just seizes up! LTSpice can be a dog sometimes
. It works OK, though, with EL34s (and different voltages, of course). It shows that there could be excessive open-loop gain and it might work better without the cathode bypass cap. on stage 1, otherwise there could be LF instability. The splitter operating points look OK with a CCS giving 6mA in the tail and 450v for B+, although 47k might be better than 56k for the plate loads.
I didn't use a cascode for the CCS in the model, I used a ring of two, which I find easier. If you do use a cascode, you're going to need to connect the base of the top transistor to something. It's probably OK having a 1k resistor (R9) to ground if you use a negative supply, as you have shown. However, without a negative supply, R9 must have a much higher value, to go between the base and B+. For safety's sake, a zener circuit would be a good idea but it seems a messy solution. So, your negative supply might be the best solution in the end, if you want to use cacoded transistors for a CCS!
HF instability could also rear its ugly head but that depends on your OP transformer more than anything else. You may have to experiment with values for Cf in approximtely the 150pF to 680pF range across Rf to get it right. Rf should be about 4.7k, I think, giving a closed loop gain of about 45. You can make it higher if you want more gain.
Your B+ at 520v seems a bit high for KT88s in AB1 UL. I would have thought 460v was more like it. The voltage for the splitter B+ at 520v seems too high as well. In any case, the splitter needs to be decoupled from the main B+ and a choke would do that nicely if you're going to drop B+ anyway; otherwise, use a 10k resistor. There has to be a decoupling cap. in there as well, whether you're using a choke or a dropping resistor: say 22uF to ground (after the dropping resistor or choke, not before it as you had last time).
The bias network on your OP tubes looks better now, although the range of adjustment you've allowed is rather a lot, from -80v to -40v (you need about -60). You can reduce the adjustment range by either increasing the resistor between the bias pot and ground (R17) or reducing the pot resistance, or both.
Don't forget to set the bias pots to the -80v end to start with, then adjust them carefully to give the required quiescent current. ~50mA seems right to start with but you need to adjust it so it sounds right to you.
I just noticed that you have no grid resistor on the first stage! The input cap should go before the volume pot, not after. It's rather high resistance for a volume control but you can get away with it because the input capacitance of a pentode is very low. Maybe shunt the slider to ground with a 470k resistor, though, so the pot won't cause a scratchy noise in the speaker if its track has imperfections (I don't trust pots!)
- I have a grid resistor in the first stage, is R1 (4K7)
- OK, I can change the pot volume to 250K. I used 500k because of the high input impedance of the tubes.
- OK for the cap before the volume pot.
I just spent some time trying to model your circuit with LTSpice , but to no avail. It seems the model I'm trying to use for KT88 is no good, it just seizes up! LTSpice can be a dog sometimes . It works OK, though, with EL34s (and different voltages, of course). It shows that there could be excessive open-loop gain and it might work better without the cathode bypass cap. on stage 1, otherwise there could be LF instability.
I've download the LTSpice but I didnt have much time to know how to use it, but I will try a little more later with ohms law and stuff 😀
Do you have any special library for tubes?
The splitter operating points look OK with a CCS giving 6mA in the tail and 450v for B+, although 47k might be better than 56k for the plate loads.
I didn't use a cascode for the CCS in the model, I used a ring of two, which I find easier. If you do use a cascode, you're going to need to connect the base of the top transistor to something. It's probably OK having a 1k resistor (R9) to ground if you use a negative supply, as you have shown. However, without a negative supply, R9 must have a much higher value, to go between the base and B+. For safety's sake, a zener circuit would be a good idea but it seems a messy solution. So, your negative supply might be the best solution in the end, if you want to use cacoded transistors for a CCS!
what does CCS means? I forgot to ask 😀
Yesterday I built the negative supply for the transistors, and something migth be wrong. I put the multimeter to measure VCE and it was like 80V !! the maxium rating for BC547 is 40v, I will try with another transistors.
HF instability could also rear its ugly head but that depends on your OP transformer more than anything else. You may have to experiment with values for Cf in approximtely the 150pF to 680pF range across Rf to get it right. Rf should be about 4.7k, I think, giving a closed loop gain of about 45. You can make it higher if you want more gain.
I'm going to use a pot and a scope to select the best R and C for the feedback with a square wave. Is that OK?
Your B+ at 520v seems a bit high for KT88s in AB1 UL. I would have thought 460v was more like it. The voltage for the splitter B+ at 520v seems too high as well. In any case, the splitter needs to be decoupled from the main B+ and a choke would do that nicely if you're going to drop B+ anyway; otherwise, use a 10k resistor. There has to be a decoupling cap. in there as well, whether you're using a choke or a dropping resistor: say 22uF to ground (after the dropping resistor or choke, not before it as you had last time).
Lucky me!!!! the man who build my psu transformer put 2 output: 200+200 and 180+180!! That will solve the problem.
The bias network on your OP tubes looks better now, although the range of adjustment you've allowed is rather a lot, from -80v to -40v (you need about -60). You can reduce the adjustment range by either increasing the resistor between the bias pot and ground (R17) or reducing the pot resistance, or both.
OK!!
Don't forget to set the bias pots to the -80v end to start with, then adjust them carefully to give the required quiescent current. ~50mA seems right to start with but you need to adjust it so it sounds right to you.
I've made that mistake once.... and my kt's went RED
I'm really carefull now each time I connect the +B to the center-tap of my output trans...
thanks a lot!!!! I really apreciate your help!!!! this forum is the best 😀
- I have a grid resistor in the first stage, is R1 (4K7)
- OK, I can change the pot volume to 250K. I used 500k because of the high input impedance of the tubes.
- OK for the cap before the volume pot.
I just spent some time trying to model your circuit with LTSpice , but to no avail. It seems the model I'm trying to use for KT88 is no good, it just seizes up! LTSpice can be a dog sometimes
. It works OK, though, with EL34s (and different voltages, of course). It shows that there could be excessive open-loop gain and it might work better without the cathode bypass cap. on stage 1, otherwise there could be LF instability. I've download the LTSpice but I didnt have much time to know how to use it, but I will try a little more later with ohms law and stuff 😀
Do you have any special library for tubes?
The splitter operating points look OK with a CCS giving 6mA in the tail and 450v for B+, although 47k might be better than 56k for the plate loads.
I didn't use a cascode for the CCS in the model, I used a ring of two, which I find easier. If you do use a cascode, you're going to need to connect the base of the top transistor to something. It's probably OK having a 1k resistor (R9) to ground if you use a negative supply, as you have shown. However, without a negative supply, R9 must have a much higher value, to go between the base and B+. For safety's sake, a zener circuit would be a good idea but it seems a messy solution. So, your negative supply might be the best solution in the end, if you want to use cacoded transistors for a CCS!
what does CCS means? I forgot to ask 😀
Yesterday I built the negative supply for the transistors, and something migth be wrong. I put the multimeter to measure VCE and it was like 80V !! the maxium rating for BC547 is 40v, I will try with another transistors.
HF instability could also rear its ugly head but that depends on your OP transformer more than anything else. You may have to experiment with values for Cf in approximtely the 150pF to 680pF range across Rf to get it right. Rf should be about 4.7k, I think, giving a closed loop gain of about 45. You can make it higher if you want more gain.
I'm going to use a pot and a scope to select the best R and C for the feedback with a square wave. Is that OK?
Your B+ at 520v seems a bit high for KT88s in AB1 UL. I would have thought 460v was more like it. The voltage for the splitter B+ at 520v seems too high as well. In any case, the splitter needs to be decoupled from the main B+ and a choke would do that nicely if you're going to drop B+ anyway; otherwise, use a 10k resistor. There has to be a decoupling cap. in there as well, whether you're using a choke or a dropping resistor: say 22uF to ground (after the dropping resistor or choke, not before it as you had last time).
Lucky me!!!! the man who build my psu transformer put 2 output: 200+200 and 180+180!! That will solve the problem.
The bias network on your OP tubes looks better now, although the range of adjustment you've allowed is rather a lot, from -80v to -40v (you need about -60). You can reduce the adjustment range by either increasing the resistor between the bias pot and ground (R17) or reducing the pot resistance, or both.
OK!!
Don't forget to set the bias pots to the -80v end to start with, then adjust them carefully to give the required quiescent current. ~50mA seems right to start with but you need to adjust it so it sounds right to you.
I've made that mistake once.... and my kt's went RED
I'm really carefull now each time I connect the +B to the center-tap of my output trans...thanks a lot!!!! I really apreciate your help!!!! this forum is the best 😀
Here is my PSU schematic
An externally hosted image should be here but it was not working when we last tested it.
Nice drawing... I wonder if your 250ma choke will be big enough for a stereo amp having 200ma at idle?? The cap in the bias supply is too big from what I've seen. .... My PS has to be different due to the fact the Dyna Mk3 power xmer will be at 535 B+ and I also will need a screen supply for pentode mode. Maybe EH KT-90s would be a better choice in my set up?? Has anybody used these in Mk3s??... So far this looks to be great project...... Regards....
crispycircuit said:
I wonder the same thing.
If its not enough, I will buy another PSU trans&choke and convert it into 2 mono-blocks.
That is not what I meant by a grid resistor - that's a grid stopper. The point is that you have to have a resistor between the grid and ground, otherwise the operating point of the tube will not be correct.
I got some tube models from the Duncan Amps website.
Nothing wrong, VCE will be high because the CCS is connected between the splitter cathodes and ground (or the negative supply if you're using that).
Using a pot won't be helpful because it's the capacitance you need to experiment with, not the resistance. You need to set the feedback resistor Rf to get the required feedback ratio and closed-loop gain. Then you need to be able to try different capacitances across it until you get an output square wave that is not round-cornered and doesn't ring.
It won't be! 250mA is only enough for one channel.
Figure on being able to draw up to 60% of the rated secondary current if you're using a cap. input smoothing filter. That limits you to 480mA. I think it'll be tight at high power and the power tranny could get very hot. Your idea of building two monoblocs might be more practical.
That is not what I meant by a grid resistor - that's a grid stopper. The point is that you have to have a resistor between the grid and ground, otherwise the operating point of the tube will not be correct.
You're right. Its not in the drawing but it is in the amp.I have a R between the stopper and ground, the value is 1M
Nothing wrong, VCE will be high because the CCS is connected between the splitter cathodes and ground (or the negative supply if you're using that).
Nothing wrong? but the datasheet in the maxium ratings says 40V vce! are you sure?
I'm not sure but what controls the current of the current-source? is the 1K resistor between the emitter and the led ?
like this
I = (Vled-Vbe)/1k
Am I right?
Using a pot won't be helpful because it's the capacitance you need to experiment with, not the resistance. You need to set the feedback resistor Rf to get the required feedback ratio and closed-loop gain. Then you need to be able to try different capacitances across it until you get an output square wave that is not round-cornered and doesn't ring.
OK!!!!
the gain is Rf/R2 ?
It won't be! 250mA is only enough for one channel.
Damn....... that sucks. I cant put a bigger choke in my chassis =(
Figure on being able to draw up to 60% of the rated secondary current if you're using a cap. input smoothing filter. That limits you to 480mA. I think it'll be tight at high power and the power tranny could get very hot. Your idea of building two monoblocs might be more practical.
That will be a bad response at low frequencies, right? I mean, to have a small-current choke.
thank you a lot. Tomorrow I will update the schematics!!!
You're right. Its not in the drawing but it is in the amp.I have a R between the stopper and ground, the value is 1M
Nothing wrong, VCE will be high because the CCS is connected between the splitter cathodes and ground (or the negative supply if you're using that).
Nothing wrong? but the datasheet in the maxium ratings says 40V vce! are you sure?
I'm not sure but what controls the current of the current-source? is the 1K resistor between the emitter and the led ?
like this
I = (Vled-Vbe)/1k
Am I right?
Using a pot won't be helpful because it's the capacitance you need to experiment with, not the resistance. You need to set the feedback resistor Rf to get the required feedback ratio and closed-loop gain. Then you need to be able to try different capacitances across it until you get an output square wave that is not round-cornered and doesn't ring.
OK!!!!
the gain is Rf/R2 ?
It won't be! 250mA is only enough for one channel.
Damn....... that sucks. I cant put a bigger choke in my chassis =(
Figure on being able to draw up to 60% of the rated secondary current if you're using a cap. input smoothing filter. That limits you to 480mA. I think it'll be tight at high power and the power tranny could get very hot. Your idea of building two monoblocs might be more practical.
That will be a bad response at low frequencies, right? I mean, to have a small-current choke.
thank you a lot. Tomorrow I will update the schematics!!!
I'm watching with great interest.... Maybe a 5 hry 500 ma would be enough. Can you parallel same value chokes for more current? I'd like to see you get your stereo version going first... Keep up the good work.......
You misunderstand me. I meant that there's nothing wrong with the voltage you measured - actually, I would expect it to be more like 95v. That doesn't mean it's OK to use a transistor rated at 40v Vce max! You definitely need to choose a transistor with a higher rating, maybe a TIP 50.
Yes, that's it. Note that junction voltages can drift with temperature, so it's important to keep the CCS circuit away from heat sources.
Roughly, yes. It's a reasonable approximation if your loop gain is high enough but it gets less accurate as the loop gain decreases. It's not all that important to be accurate here, since you can play with the NFB to get the gain you need.
According to the Genalex data sheet for KT88 in UL PP Class AB1, with 460v on plate and g2, fixed bias at -69v, 4k plate-to-plate load and 40% UL taps, the quiescent plate+screen current per tube should be 50mA. At full signal, for 70w output, the plate+screen current per tube could reach 140mA per tube. That is a total of 280mA, plus about 8mA for the other stages, per channel.
Although you probably won't run it flat out like that, it will still get close to 250mA. Since there are two channels , you really need separate supplies or a heavier current choke. It only needs to be 3 Henries but it must be able to handle 600mA. If it's under-specified, it will saturate and lose inductance at high current. PS regulation will suffer and the voltage will sag, causing poor LF performance, as you say.
Hi crispycircuit,
Your project sounds interesting too! I don't know what is the plate-to-plate impedance of the Dynaco Mk III OP transformer, but I expect it will probably do the job for pentode mode. You just need to insulate the unused primary taps.
I looked at the Genalex data sheet, which shows figures for KT88 in pentode mode PP Class AB1. With plate voltage at 560v, grid 2 at 300v, 4.5k load plate-to-plate and -34v bias on grid 1, it claims to be capable of giving 100w output. Quiescent plate current is quoted as 60mA per tube and 140mA at full signal.
As I'm sure you're aware, the grid 2 voltage needs to be kept constant at 300v for good quality sound. It's also essential to use grid stoppers on grid 1 and grid 2 or you will get parasitics with that tube!
One of the regulators mentioned in other threads on this site would be suitable for supplying grid 2. Apart from that and the different bias setting, the circuit being developed by Zafira would probably suit your needs (a hell of a lot better than the Dynaco 😀 )
Your project sounds interesting too! I don't know what is the plate-to-plate impedance of the Dynaco Mk III OP transformer, but I expect it will probably do the job for pentode mode. You just need to insulate the unused primary taps.
I looked at the Genalex data sheet, which shows figures for KT88 in pentode mode PP Class AB1. With plate voltage at 560v, grid 2 at 300v, 4.5k load plate-to-plate and -34v bias on grid 1, it claims to be capable of giving 100w output. Quiescent plate current is quoted as 60mA per tube and 140mA at full signal.
As I'm sure you're aware, the grid 2 voltage needs to be kept constant at 300v for good quality sound. It's also essential to use grid stoppers on grid 1 and grid 2 or you will get parasitics with that tube!
One of the regulators mentioned in other threads on this site would be suitable for supplying grid 2. Apart from that and the different bias setting, the circuit being developed by Zafira would probably suit your needs (a hell of a lot better than the Dynaco 😀 )
Ray thanks for your reply. The Dyna Mk3 output is 4300ct. The Mk3 power is 820ct at 200ma.... What current production tube would you choose?...
For the output tube grid 1 is 2.2K good? and grid 2 a 470 ohm? With regulation do you need Grid 2 stopper resistor?
Do you have a favorite regulator for my 300vdc supply (screen). And on the B+ filtering, do you like large capasitance and would a choke improve the supply some??? Choke or no choke?? It looks more and more like I'll have to buy a larger chassis and skip the original Mk3 chassis.... Ok, that's my questions for the day.....
For the output tube grid 1 is 2.2K good? and grid 2 a 470 ohm? With regulation do you need Grid 2 stopper resistor?
Do you have a favorite regulator for my 300vdc supply (screen). And on the B+ filtering, do you like large capasitance and would a choke improve the supply some??? Choke or no choke?? It looks more and more like I'll have to buy a larger chassis and skip the original Mk3 chassis.... Ok, that's my questions for the day.....
Sounds good!
Voltage seems a bit low for pentode mode, from the data sheet I've seen. As for the current rating, that's much lower than I would have expected, unless that's the rating for the whole winding. You can only get about 60% of the rated current with a cap. input filter and at full signal you're possibly looking at ~140mA each for the OP tubes.
It might work fine but with available OP power less than the data sheet's theoretical 100w. With that OP tranny, you can always change to UL if you don't like the results in pentode mode! 🙂
I'd rather leave that question for someone else with enough experience of the differemt brands of KT88.
2.2k grid 1 stopper is OK. Yes, you do need a stopper on grid 2 of each tube, even if you are using regulation. 150 ohm is enough; you don't want to make it too big, or you'll lose the benefits of regulation. BTW, stoppers need to be rated at around 2w, to withstand heating from the grid pins, and must NOT be wirewound. The resistor needs to be vey close indeed to the pin, so you need to trim the lead going to the grid pin as short as possible.
I've been asking questions about this myself recently. From the replies I got, I reckon SY's floating LM317 is probably as good as any - see this thread
I like to have plenty of (good quality) capacitance (like 470uF) as the reservoir cap, i.e. on the output from the filter. The first (smoothing) cap I believe should be much smaller, say 47uF.
I like having a choke in the filter, inductance depending on the current to be supplied. About 5H should be suitable for your needs. Make sure it's rated for enough current, so the core doesn't saturate. Also, make sure the choke is rated for the voltage it will be carrying - this is an insulation/safety thing. If you have any doubt, it's advisable to connect the choke in the negative rather than the positive side of the B+ supply.
Mine, too! As always, if I've given any wrong advice I hope someone will jump in and correct me.
Ray,
Thanks, your getting me closer to finalizing a design... I have to get new chokes and I'll check the voltage rating... Does anybody sell regulator boards? I did build the Last PAS in Glass Audio years ago and I think they sold one??
Anybody know of a source for 5 watt zeners? If you find any more regulator designs for my purpose I'm still collecting schematics... Also you said "good quality filter caps". Your thoughts? Computer grade, JJ's, Solen, Film or electro's...... If I'm gonna buy them, it may as well be something that will get the job done correctly....... I better start returning my beer bottles!
Thanks, your getting me closer to finalizing a design... I have to get new chokes and I'll check the voltage rating... Does anybody sell regulator boards? I did build the Last PAS in Glass Audio years ago and I think they sold one??
Anybody know of a source for 5 watt zeners? If you find any more regulator designs for my purpose I'm still collecting schematics... Also you said "good quality filter caps". Your thoughts? Computer grade, JJ's, Solen, Film or electro's...... If I'm gonna buy them, it may as well be something that will get the job done correctly....... I better start returning my beer bottles!
Hello everybody. I've been busy with some test at the college. I've finished so I will continue with the project. Today I will build the new CCS and tomorrow I will post the changes.
Thanks to all the people interested in my project.
Jorge.
Thanks to all the people interested in my project.
Jorge.
I built the CCS with 2xBD139 and 2 diodes 1n4002
vce1=18.5v
vce2=54.6v
it seems that is working OK 😀
An externally hosted image should be here but it was not working when we last tested it.
vce1=18.5v
vce2=54.6v
it seems that is working OK 😀
Thanks for the update... Looking forward to your results. ... Are you going to use a HV regulator for the EF86 supply?? Would this improve the sound?? ... Or should I build a seperate supply for the pentode (6550) screen grids and run the EF86 off that?? Screens need to be at 300 volts....
crispycircuit said:
I will not use regulators unless I found some noise or problems, because I dont have room in my chassis. Maybe if I build 2 mono-blocks because of my low-amperage choke. I dont know if that would improve the sound. This is my first tube amp so I cant answer that questions.
Wait for ray_moth and lets see his opinion 😀
rigth now I'm modifiying the schematic to post it.
thanks!!
Jorge.
I'm in the process of finding decent chassis for my mono's. I like some room to work with. This project is going to be more expensive that planned but really fits my needs... Your help is wonderful......
- Status
- Not open for further replies.