Shoog,
You have to derate the voltage on that 60Hz transformer you are planning for an interstage if you want response below 60Hz. Dropping the winding voltages to 1/2 will get you to 30 Hz. Dropping to 1/3rd gives 20 Hz.
I think you can do this with a 6SN7GTB on a separate HV supply using the first stage as input gain and the second stage as a concertina splitter and driver.
You have to derate the voltage on that 60Hz transformer you are planning for an interstage if you want response below 60Hz. Dropping the winding voltages to 1/2 will get you to 30 Hz. Dropping to 1/3rd gives 20 Hz.
I think you can do this with a 6SN7GTB on a separate HV supply using the first stage as input gain and the second stage as a concertina splitter and driver.
The transformers will be 50hz rated and are 50VA. Since I will be driving them with about an 18V signal off the 5687's with only about 15-20mA per transformer I don't think there will be issues with saturation. There may be issues with HF roll of due to the high inductance of the driver side - though I doubt it.
I don't think anyone has documented trying this approach so I am voyaging into uncharted territory. It may or may not work, but I don't think I will lose much by trying (apart from the cost of 2 transformers). It has an elegant simplicity which appeals to me, though of course it may be to simple. I am more concerned with resonant tanks than anything else. sizing the coupling cap is a real suck it and see scenario.
Shoog
I don't think anyone has documented trying this approach so I am voyaging into uncharted territory. It may or may not work, but I don't think I will lose much by trying (apart from the cost of 2 transformers). It has an elegant simplicity which appeals to me, though of course it may be to simple. I am more concerned with resonant tanks than anything else. sizing the coupling cap is a real suck it and see scenario.
Shoog
OK, sounds like you have this covered. I think the Z of a 50 volt winding will be totally too low for use as an IT if plate coupled but you are right, using an IT does end up with a simpler looking schematic than other methods.
The beauty of DIY is that you can find this stuff out yourself and get a practical and valuable education along with the fun, all at the same time. In rare instances you even save money. .)
The beauty of DIY is that you can find this stuff out yourself and get a practical and valuable education along with the fun, all at the same time. In rare instances you even save money. .)
That won't solve your Z problem. Your efficiency will be in the toilet, and the distortion up. Don't try to sell ME on it though. You are the one this is for. Go for it. You have uncharted territory (to you) to explore.
If it was my house, I'd be trying the 6SN7 as I suggested above without iron.
If it was my house, I'd be trying the 6SN7 as I suggested above without iron.
"That won't solve your Z problem. Your efficiency will be in the toilet, and the distortion up. Don't try to sell ME on it though. You are the one this is for. Go for it. You have uncharted territory (to you) to explore."
Excuse my ignorance if I am wrong, but is my thinking on this correct.
If I go for the maximum grid circuit resistance of 1meg ohm for a 6080, and I am using a 1:2 ratio I get a reflected resistance of 250K. If this is in parallel with a CCS I should be loading the 5687 by about 240K which is a good load is it not.
If my logic on this one is wrong can you explain to me how, so that I can learn.
Shoog
Excuse my ignorance if I am wrong, but is my thinking on this correct.
If I go for the maximum grid circuit resistance of 1meg ohm for a 6080, and I am using a 1:2 ratio I get a reflected resistance of 250K. If this is in parallel with a CCS I should be loading the 5687 by about 240K which is a good load is it not.
If my logic on this one is wrong can you explain to me how, so that I can learn.
Shoog
Shoog said:
If you need a mathemagical explanation I'm not your best tutor. I do a lot of thinking like this with my hands which I can use to pick up my handy reactance chart. It works for me. For any given frequency as you increase the number of turns and increase L, you also increase the AC impedance. For a fixed number of turns on a given core, as you lower frequency the Z will drop. The plate circuit of your driver stage wants a pretty high plate load Z, otherwise the dinky amount of plate current available won't be capable of creating a sufficient voltage drop across the coil.
Forgetting about frequency or any loading, if you have a plate swing of 100 volts at 10 mA, that primary winding has to look like 10K ohms. Now consider bass frequencies which require the most L for a given current. Looking at my handy reactance chart I see that 10K Z= a 30 Hy coil at 60 Hz. To swing 100 volts at 20 Hz from a 10 mA source would require about 85 Hy's.
As a power xfmer, if your 50 volt winding is rated at 50 watts at 50 Hz that is 1 amp. According to Ohms Law (not R's Law) that is a Z of 50 ohms. According to my handy reactance chart that coil is about 0.16 Henries. You are way on the short side of not enough L.
By cap coupling this winding to the driver tube plate you will get whatever voltage across it that can be developed with 10 mA. At 50Hz that would be about 0.5 volts. The waveform will be extremely distorted. Now add secondary windings and loading and this gets even worse. Try to reduce the frequency to 20 Hz and it gets even worse again.
You might be able to use this xfmer as an interstage but it would have to be cathode driven by a tube with a low output Z. Maybe another 6080 with sections parallelled.
Absolutely the same applies to output xfmers, both directly in the plate circuit DC path or cap coupled only to the AC component. With a power stage however the plate Z is much lower, so lower L may be tolerated.
My PSET 6080 amp design has a really low plate Z with 8 triodes in parallel. I discovered that a certain PS toroid I found at a surplus store was great with two (120 volt primaries in series, low voltage secondaries in parallel) as an output xfmer for this amp, so I bought up as many as I could, enough to make a few of these amps eventually. I use two ~80 VA cores (~160 VA) at 60 Hz to handle 20 Hz at about 20+ watts with negligible waveform distortion.
My PSET 6080 amp design has a really low plate Z with 8 triodes in parallel. I discovered that a certain PS toroid I found at a surplus store was great with two (120 volt primaries in series, low voltage secondaries in parallel) as an output xfmer for this amp, so I bought up as many as I could, enough to make a few of these amps eventually. I use two ~80 VA cores (~160 VA) at 60 Hz to handle 20 Hz at about 20+ watts with negligible waveform distortion.
I put my hands up and declare, I am enlightened of my ignorance. So the games up and the idea has to be abandoned.
But I am not keen on spending €200.00 on interstage transformers, and I'am not keen on using tube based phase splitting. I am not ready to abandon the idea yet.
But wait a minuite, there is a factor of any parafeed circuit which can be turned to our advantage. A parafeed circuit has a tuned resonance, usually we tune that resonance to be as near to DC as possable, but it is possable to tune it up and use it as a bass boost. This is quite common practice in parafeed output stages. The fequency of the resonance is dependent on the primary inductance, the parafeed capacitance and the output impedence of the driver circuit.
The primary inductance is fixed.
The capacitance can be varied, but is a pain to fiddle with and it might require a electo capacitor.
The output impedence of the driver is seemingly fixed, but on closer examination it is possable to varie it by using a two stage driver and applying partial feedback between the plates. So starting with a ECL82, we can apply the plate to plate feedback with a 200K pot, which would allow us to tune the resonance by ear to somewhere around 10-20Hz.
Sounds like a possablity. Care to express an opinion.
I just went back and compared my Partial Feedback parafeed SET amp to my best gainclone, and did notice that the bass was definately rolled off. So I applied the above principle and reduced the 190K Plate to Plate feedback to 115K. Result and significant improvement in frequencies bellow 50hZ.
Unfortunately this is not an option I have with getting the mains toroidal output transformers in my proposed PP amp working. I may just have to look for a 1K2 PP output transformer.
Shoog
But I am not keen on spending €200.00 on interstage transformers, and I'am not keen on using tube based phase splitting. I am not ready to abandon the idea yet.
But wait a minuite, there is a factor of any parafeed circuit which can be turned to our advantage. A parafeed circuit has a tuned resonance, usually we tune that resonance to be as near to DC as possable, but it is possable to tune it up and use it as a bass boost. This is quite common practice in parafeed output stages. The fequency of the resonance is dependent on the primary inductance, the parafeed capacitance and the output impedence of the driver circuit.
The primary inductance is fixed.
The capacitance can be varied, but is a pain to fiddle with and it might require a electo capacitor.
The output impedence of the driver is seemingly fixed, but on closer examination it is possable to varie it by using a two stage driver and applying partial feedback between the plates. So starting with a ECL82, we can apply the plate to plate feedback with a 200K pot, which would allow us to tune the resonance by ear to somewhere around 10-20Hz.
Sounds like a possablity. Care to express an opinion.
I just went back and compared my Partial Feedback parafeed SET amp to my best gainclone, and did notice that the bass was definately rolled off. So I applied the above principle and reduced the 190K Plate to Plate feedback to 115K. Result and significant improvement in frequencies bellow 50hZ.
Unfortunately this is not an option I have with getting the mains toroidal output transformers in my proposed PP amp working. I may just have to look for a 1K2 PP output transformer.
Shoog
I decided to try an experiment to see how a transformer would perform. I put a 10V signal into a little 30VA RS toroidal with a ration of 24V:110V and applied the signal to be stepped up. I loaded the secondary with a 470K resistor.
I swept a signal down from 20Khz down to about 10hz. There was a slight fall in response down from the KHZ range, but no significant drop off below 100hz. The sine wave came through clean.
I also applied a square wave signal and got sever spiking and ringing, in the order of 3 times the main signal. This is to be expected when a secondary is so lightly loaded. Hopefully the grid capacitance and a bit of capacitance over the secondary would calm this.
I also tried a much larger toroidal with a ratio of 48V:240V. This showed much more pronounced drop off in the bass, with a -3db point at about 20hz. The sine wave signal was heavily distorted below 20hz, and lightly distorted below 50hz.
Anyone care to duplicate my experiment to confirm my findings.
Just my finding. Makes me think the little toroidals will work if the ringing can be calmed. Anyone else got any useful observations.
Shoog
I swept a signal down from 20Khz down to about 10hz. There was a slight fall in response down from the KHZ range, but no significant drop off below 100hz. The sine wave came through clean.
I also applied a square wave signal and got sever spiking and ringing, in the order of 3 times the main signal. This is to be expected when a secondary is so lightly loaded. Hopefully the grid capacitance and a bit of capacitance over the secondary would calm this.
I also tried a much larger toroidal with a ratio of 48V:240V. This showed much more pronounced drop off in the bass, with a -3db point at about 20hz. The sine wave signal was heavily distorted below 20hz, and lightly distorted below 50hz.
Anyone care to duplicate my experiment to confirm my findings.
Just my finding. Makes me think the little toroidals will work if the ringing can be calmed. Anyone else got any useful observations.
Shoog
I went back and checked my figures. The input voltage was nearer to 6V. Also I have just learned that the -3db point is at 0.707 x the midrange voltage, and not half as I though. I therefore have to correct myself and say that the -3db point is at 10hz. This seems comparable to most commercial interstage transformers.
I will get some 110+110V:55+55V tranformers. This will allow me to configure them as either a 1:2+2 voltage stepup which should give me an input sensitivity of about 1V to drive the 6080's fully using a 5687. This will only put about 18V onto the primary which is well bellow their 55V rating.
Or I can configure them as a 1:1+1 which should give me an input sensitivity of 2V to drive the 6080's to full ouput. Again I will only be putting 36V onto the primary which is 1/3 of there rated voltage.
If this does show to much low frequency roll off I will replace the 5687 with ECL82's with plate to plate feedback so that I can tune the resonance as I suggested before.
I am prepared to take the risk that this won't work out, for an outlay of only €35.00 on transformers. I will do measurements when I do, so that I can establish whats going on. It might be a few months before I have any results. I will build it with mains toroidal outputs and if they show to much low frequency roll of I will look for better.
Shoog
I will get some 110+110V:55+55V tranformers. This will allow me to configure them as either a 1:2+2 voltage stepup which should give me an input sensitivity of about 1V to drive the 6080's fully using a 5687. This will only put about 18V onto the primary which is well bellow their 55V rating.
Or I can configure them as a 1:1+1 which should give me an input sensitivity of 2V to drive the 6080's to full ouput. Again I will only be putting 36V onto the primary which is 1/3 of there rated voltage.
If this does show to much low frequency roll off I will replace the 5687 with ECL82's with plate to plate feedback so that I can tune the resonance as I suggested before.
I am prepared to take the risk that this won't work out, for an outlay of only €35.00 on transformers. I will do measurements when I do, so that I can establish whats going on. It might be a few months before I have any results. I will build it with mains toroidal outputs and if they show to much low frequency roll of I will look for better.
Shoog
Hi there,
I just thought I would give a status report on how the project has been going.
I have built the power supply. I can report that a microwave oven transformer performs very well as a filter choke and doesn't seem to saturate at up to 1/2 amp of DC current. I have a final +B of about 115V which is a tad shy of what I would want. The choke has a DCR of 100R so it drops about 50V all by itself. Unfortunately my DMM has died and so I am having to do readings from my scope.
I built the cathode bias network. The 6AS7 is very variable, I am using individual resistances for each cathode with a paralleled 1K plus a 1K pot for bias adjustment, this gives me a an adjustment range of about 280R to 330R, which is barely enough to get currents balanced. May need a rethink on this.
Got the toroidals for the interstage today. They are 0-55V,0-55V:115V,111V at 50VA. I used my signal generator to do some basic tests,with a 6Vpp input signal (unknown current). With the secondaries in series to give a 1:1+1. I tried the "created" primary wired in one direction and got terrible results. It rolled off below 200hz and rolled off at the top end at about 30khz, before rising again after 40khz. There also seemed significant ringing. Not good at all I though.
I then tried reversing the primary connection. This produced markedly different results. There was no real low end roll off down to 10hz, at the top end the -3db point was at about 20khz and it continued rolling off from there without any rise. With a sine wave the signal came through increadably cleanly. With a square wave it stayed good down to about 200hz and then went somewhat triangular. At the top end the leading edge was rolled off after about 10khz (no ringing at all surprisingly). Of course I will be putting in at least 6 times this amount of voltage in the final design, so things may not be quite so rosy, but things look very promising so far. Also there was no load on the secondary, but also I would say that the signal generator would have nothing like the current output of my 5687 driver.
Will keep you all posted.
Shoog
I just thought I would give a status report on how the project has been going.
I have built the power supply. I can report that a microwave oven transformer performs very well as a filter choke and doesn't seem to saturate at up to 1/2 amp of DC current. I have a final +B of about 115V which is a tad shy of what I would want. The choke has a DCR of 100R so it drops about 50V all by itself. Unfortunately my DMM has died and so I am having to do readings from my scope.
I built the cathode bias network. The 6AS7 is very variable, I am using individual resistances for each cathode with a paralleled 1K plus a 1K pot for bias adjustment, this gives me a an adjustment range of about 280R to 330R, which is barely enough to get currents balanced. May need a rethink on this.
Got the toroidals for the interstage today. They are 0-55V,0-55V:115V,111V at 50VA. I used my signal generator to do some basic tests,with a 6Vpp input signal (unknown current). With the secondaries in series to give a 1:1+1. I tried the "created" primary wired in one direction and got terrible results. It rolled off below 200hz and rolled off at the top end at about 30khz, before rising again after 40khz. There also seemed significant ringing. Not good at all I though.
I then tried reversing the primary connection. This produced markedly different results. There was no real low end roll off down to 10hz, at the top end the -3db point was at about 20khz and it continued rolling off from there without any rise. With a sine wave the signal came through increadably cleanly. With a square wave it stayed good down to about 200hz and then went somewhat triangular. At the top end the leading edge was rolled off after about 10khz (no ringing at all surprisingly). Of course I will be putting in at least 6 times this amount of voltage in the final design, so things may not be quite so rosy, but things look very promising so far. Also there was no load on the secondary, but also I would say that the signal generator would have nothing like the current output of my 5687 driver.
Will keep you all posted.
Shoog
Progress report.
Got one channel lashed up on the breadboard. Had the radio playing through it. There are a few problems though. Seems not to have anywhere near enough gain (Ie unity voltage gain). Also the interstage transformer is performing reasonably well, but has a dip in response at about 10KHz and rises after that. This is not good.
The main issue I am having though is with my CCS loaded 5687 drivers. I am using SY's jiggy CCS. I had it set up originally with 125V at the top of the CCS, giving me a plate voltage of 100V and a bias point of 3.7V and about 9mA of current. This caused sever clipping with anything more than about 0.5Vpp of input. So I tried taking the driver supply off before the PS choke, which gave me about 165V on the top of the CCS, the plate voltage has now risen up to 135V and the clipping is still there. Don't know how to force the CCS to consume more voltage. If I can't get this working I will have to try resistive loading.
Shoog
Got one channel lashed up on the breadboard. Had the radio playing through it. There are a few problems though. Seems not to have anywhere near enough gain (Ie unity voltage gain). Also the interstage transformer is performing reasonably well, but has a dip in response at about 10KHz and rises after that. This is not good.
The main issue I am having though is with my CCS loaded 5687 drivers. I am using SY's jiggy CCS. I had it set up originally with 125V at the top of the CCS, giving me a plate voltage of 100V and a bias point of 3.7V and about 9mA of current. This caused sever clipping with anything more than about 0.5Vpp of input. So I tried taking the driver supply off before the PS choke, which gave me about 165V on the top of the CCS, the plate voltage has now risen up to 135V and the clipping is still there. Don't know how to force the CCS to consume more voltage. If I can't get this working I will have to try resistive loading.
Shoog
Hi Shoog
If I understood Morgan Jones book (valve amplifiers, 3rd edition, page 137) well you can set the anode voltage of the tube by looking at the plate curves (pretty logic, actually).
For an anode voltage of 100V at a 9mA draw you need the grid to be biased at a little less than -4V (something around -4,2V). MJ uses a zenerdiode to accomplish it. What voltage are you measuring at the cathode?
If I am wrong, there will be sufficient people around to correct me.
Erik
If I understood Morgan Jones book (valve amplifiers, 3rd edition, page 137) well you can set the anode voltage of the tube by looking at the plate curves (pretty logic, actually).
For an anode voltage of 100V at a 9mA draw you need the grid to be biased at a little less than -4V (something around -4,2V). MJ uses a zenerdiode to accomplish it. What voltage are you measuring at the cathode?
If I am wrong, there will be sufficient people around to correct me.
Erik
There is a just over 4V on the cathode. The CCS just doesn't seem to want to pass that little current so, whatever setting I have on the CCS's adjustment pot the voltage stays stubbornly at about 140V(or higher). I have since discovered a few wiring errors in the CCS, but correcting them still doesn't change this 140V fact. 25V seems to be the most I can get it to drop.
Shoog
Shoog
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.