... why not build one with "proper" gain devices, a.k.a tubes?
The most natural topology seems to be a pentode output stage (preferably SE) with little or no NFB. The lack of NFB calls for a linear pentode, which puts quite a few of our common suspects off the list.
Power requirements are in my case very moderate (my Fostex FE107E´s in BL horns are happy with my <1W PCL82 amp), so there´s no need to dig up some 813´s or something.
Most people agrees that DH tubes sounds better than IDH´s, so what we´re looking for is a small and linear direct heated pentode.
The americans have their 47´s, a very rare tube here in Sweden. On the other hand we have the AL1 which is pretty much the same thing (well, they play in the same division at least).
http://frank.pocnet.net/sheets/066/a/AL1.pdf
The datasheet calls for 7k plate load, which seems to be an odd number when it comes to availability of OPT´s. What I have in my junkbox is a pair of 5,6k/8R OPTs of reasonable quality, but if I´ve understood John Swensons theories about pentode transconductance amps the lower impedance should be a good thing.
Well, so far so good, right?
AL1 requires about +-15V input swing so a driver stage is of course needed. My first idea was to use a simple triode wired EF86, mostly because I have a source for cheap, high quality ones. It would probably be a good choice, but how boring wouldn´t it be???
A DHP transconductance output stage calls for an equally weird driver stage, so what about another DHP?
"My" tube dealer has a few thousands of small (and I mean small) DHPs intended for battery equipment.
For example: 1S4, 3A4, 3V4, 3Q4 and so on. Price? Next to nothing.
Some of them measures good as triodes, but the gain would be way too low. The interesting thing is that some of them actually measures good as pentodes too, you see where I´m heading?
Heating all these DHP´s with clean DC will be a mess, but these small critters draws something like 100mA @ 1,4V so at least we´re not talking any big heavy stuff...
Well well well, this amp might not be the answer to all our prayers but my speakers seems to like a bit of Zout and building a beast like this might be the adventure of a lifetime.
As far as I can see I have everything within reach, I´ve got power tubes, sockets, OPT´s, PT (or at least I know how to wind one), PSU choke and much of the miscellanous stuff needed.
Gents, are you with me or against me??
/Daniel
The most natural topology seems to be a pentode output stage (preferably SE) with little or no NFB. The lack of NFB calls for a linear pentode, which puts quite a few of our common suspects off the list.
Power requirements are in my case very moderate (my Fostex FE107E´s in BL horns are happy with my <1W PCL82 amp), so there´s no need to dig up some 813´s or something.
Most people agrees that DH tubes sounds better than IDH´s, so what we´re looking for is a small and linear direct heated pentode.
The americans have their 47´s, a very rare tube here in Sweden. On the other hand we have the AL1 which is pretty much the same thing (well, they play in the same division at least).
http://frank.pocnet.net/sheets/066/a/AL1.pdf
The datasheet calls for 7k plate load, which seems to be an odd number when it comes to availability of OPT´s. What I have in my junkbox is a pair of 5,6k/8R OPTs of reasonable quality, but if I´ve understood John Swensons theories about pentode transconductance amps the lower impedance should be a good thing.
Well, so far so good, right?
AL1 requires about +-15V input swing so a driver stage is of course needed. My first idea was to use a simple triode wired EF86, mostly because I have a source for cheap, high quality ones. It would probably be a good choice, but how boring wouldn´t it be???
A DHP transconductance output stage calls for an equally weird driver stage, so what about another DHP?
"My" tube dealer has a few thousands of small (and I mean small) DHPs intended for battery equipment.
For example: 1S4, 3A4, 3V4, 3Q4 and so on. Price? Next to nothing.
Some of them measures good as triodes, but the gain would be way too low. The interesting thing is that some of them actually measures good as pentodes too, you see where I´m heading?
Heating all these DHP´s with clean DC will be a mess, but these small critters draws something like 100mA @ 1,4V so at least we´re not talking any big heavy stuff...
Well well well, this amp might not be the answer to all our prayers but my speakers seems to like a bit of Zout and building a beast like this might be the adventure of a lifetime.
As far as I can see I have everything within reach, I´ve got power tubes, sockets, OPT´s, PT (or at least I know how to wind one), PSU choke and much of the miscellanous stuff needed.
Gents, are you with me or against me??
/Daniel
I've had a tube based current source amp driving my Lowther speakers for a month or so. I've been working with beam deflection tubes this summer so I used a 6DJH as the "power" tube (its already a pseudo differential amp in one tube) in a PP circuit. The 6DJH is a very linear transconductance amp as long as you don't over drive it.
I was supposed to get over a watt out of this but it turned out to be only 1/4 watt because of the low mu of these devices, I couldn't get enough gain out of the tube so the maximum input voltage didn't produced a high enough voltage swing at the low current levels.
This amp sounds incredible with the lowthers. To me this is the best sound I have ever heard from a stereo system, it is simply so real its scary. The only problem is that it does not go loud enough (only 1/4 watt) on really loud crescendos on clasical music that has not been compressed to death.
Right now I'm working on a "high power" output stage using the concepts in Smoking Amps thread on current mirrors. These use a tube diode as a nonlinear load to a transconductance amp, this nonlinear voltage then drives a pentode, the two curves canceling each other out, giving you a very linear transconductance amp.
Smoking Amp used strings of signal diodes to get the curves right, I've been playing with using triodes so I can adjust the curve using the grid voltage. Right now I'm using an ECL85 because it has both a triode and a pentode in one envelope. I'm just starting to test this out to see if I can really get the curves to match. If this works out I'll probably build a PP amp with the ECL85s driven by a 6DJH which does a great job at being both a transconductance amp and a phase splitter. This should give me a real 1.5 watts which should be fine for my application. IF I need more I'm planning a 6L6 based PP current mirror design.
One thing you've probably already figured out when making transconductance amps is that you have to make sure you have enough voltage complience in the amp to handle the highest impedance of the load. In general this means we have to use OPTs with a lower ratio than we normally would use for a tube amp. We usually choose the OPT so the maximum voltage range of the output stage and the maximum current are developed with an 8 ohm load. (or something similar). If you then use a transconductance amp driving a speaker that has a max impedance of 50 ohms, your transconductance amp is going to run out of voltage trying to push its specified current through the load. The solution here of course is to use an OPT with a lower ratio so the maximum power of the amp occurs at the highest impedance load of the speaker. Of course this means you won't get max power at 8 ohms, but who cares.
This is a really fun area to play in right now since it hasn't been "done to death", its kind of like thinking sideways when dealing with current amps.
John S.
I was supposed to get over a watt out of this but it turned out to be only 1/4 watt because of the low mu of these devices, I couldn't get enough gain out of the tube so the maximum input voltage didn't produced a high enough voltage swing at the low current levels.
This amp sounds incredible with the lowthers. To me this is the best sound I have ever heard from a stereo system, it is simply so real its scary. The only problem is that it does not go loud enough (only 1/4 watt) on really loud crescendos on clasical music that has not been compressed to death.
Right now I'm working on a "high power" output stage using the concepts in Smoking Amps thread on current mirrors. These use a tube diode as a nonlinear load to a transconductance amp, this nonlinear voltage then drives a pentode, the two curves canceling each other out, giving you a very linear transconductance amp.
Smoking Amp used strings of signal diodes to get the curves right, I've been playing with using triodes so I can adjust the curve using the grid voltage. Right now I'm using an ECL85 because it has both a triode and a pentode in one envelope. I'm just starting to test this out to see if I can really get the curves to match. If this works out I'll probably build a PP amp with the ECL85s driven by a 6DJH which does a great job at being both a transconductance amp and a phase splitter. This should give me a real 1.5 watts which should be fine for my application. IF I need more I'm planning a 6L6 based PP current mirror design.
One thing you've probably already figured out when making transconductance amps is that you have to make sure you have enough voltage complience in the amp to handle the highest impedance of the load. In general this means we have to use OPTs with a lower ratio than we normally would use for a tube amp. We usually choose the OPT so the maximum voltage range of the output stage and the maximum current are developed with an 8 ohm load. (or something similar). If you then use a transconductance amp driving a speaker that has a max impedance of 50 ohms, your transconductance amp is going to run out of voltage trying to push its specified current through the load. The solution here of course is to use an OPT with a lower ratio so the maximum power of the amp occurs at the highest impedance load of the speaker. Of course this means you won't get max power at 8 ohms, but who cares.
This is a really fun area to play in right now since it hasn't been "done to death", its kind of like thinking sideways when dealing with current amps.
John S.
Hmm, maybe I should aim for an ever lower (or is it higher?) load than 5,6k?
My OPT´s have four identical primary sections that each has a ratio of 8,7:1, which means I can get 600R:8R, 2,4k:8R, 5,6k:8R or 9,6k:8R. I guess that leaves room for some experiments.
BTW: What´s your opinions about DHP input stages?
My OPT´s have four identical primary sections that each has a ratio of 8,7:1, which means I can get 600R:8R, 2,4k:8R, 5,6k:8R or 9,6k:8R. I guess that leaves room for some experiments.
BTW: What´s your opinions about DHP input stages?
Use a lowish turns ratio, i.e., the entire secondary. Pentodes are pretty happy in transconductance mode, and you want to take advantage of all the primary/secondary coupling you can get, as well as a healthy compliance voltage. Optimum primary impedance will depend on the output tube- as you go lower, the compliance increases, but max current decreases. It's a trade-off that can be best side-stepped by proper output tube choice.
As far as input stages, I would still maintain that triodes are the surest route to linear voltage amplification. I'm not convinced that getting rid of the cathode is a particularly good idea. FWIW, the input stage in my amp uses a 5691, direct coupled to a 5692 driver. No pentodes until the output.
As far as input stages, I would still maintain that triodes are the surest route to linear voltage amplification. I'm not convinced that getting rid of the cathode is a particularly good idea. FWIW, the input stage in my amp uses a 5691, direct coupled to a 5692 driver. No pentodes until the output.
Maybe the 2,4k:8R connection would be best then. If so, I can wire the primary sections up in parallel pairs which then are connected in series = all sections used.
I agree that triodes might be a better choice for input stage, but the "coolness" of DHP input stages cannot be denied
I think I´ll try them DHPs and if they fail I can always change to triodes later
I agree that triodes might be a better choice for input stage, but the "coolness" of DHP input stages cannot be denied
I think I´ll try them DHPs and if they fail I can always change to triodes later
Hey transconductance gurus !
Please apologies for my ignorance !
The simplest statement I've read about Transconductance amps is "voltage controlled current source". Is that true ?
I can understand how a penthode is nearly that, but ... shortly ... how does this current swing feed the loudspeaker ?
From that thread it appears that trannies are used, so what is "different" in this technology ?
Could someone points me to an interresting link ?
PSE, simple maths
TIA
Yves.
Please apologies for my ignorance !
The simplest statement I've read about Transconductance amps is "voltage controlled current source". Is that true ?
I can understand how a penthode is nearly that, but ... shortly ... how does this current swing feed the loudspeaker ?
From that thread it appears that trannies are used, so what is "different" in this technology ?
Could someone points me to an interresting link ?
PSE, simple maths
TIA
Yves.
A further advantage of using a low primary impedence OPT (ca. 2K) is that at that load nearly all of the distortion is 2nd harmonic for a pentode, i.e. 3rd harmonics are very low, so a push-pull configuration can give very low distortion without feedback, on par with a triode amp.
I am currently working on one that uses a pair of 7119's to drive 6AR6/6098 push-pull (step-up input phase-splitting transformer LL1676) with a 2.6K or so OPT. Any comments? What kind of bias points are you guys using? Class A seems to be the best choice with low opt turns ratios and usually efficient speakers.
John
Steve: Resonance is the point where you need the least current for the greatest motion. Besides which, I'm an evangelist for not using tube amps at low frequencies, so it's sort of moot around my house.
If you still worry about that impedance peak, it's easy to drop it a few notches with a paralleled series-notch dual. Nelson Pass suggests some values for some suitable drivers in the FirstWatt manual.
Yvesm, do a search for Mills and Hawksford's JAES paper on distortion reduction via current drive. They pound out most of the basics there.
John, I'm going the other way, AB and maybe B.
If you still worry about that impedance peak, it's easy to drop it a few notches with a paralleled series-notch dual. Nelson Pass suggests some values for some suitable drivers in the FirstWatt manual.
Yvesm, do a search for Mills and Hawksford's JAES paper on distortion reduction via current drive. They pound out most of the basics there.
John, I'm going the other way, AB and maybe B.
SY said:
Um, well yeah, when you're driving the speaker with a voltage source resonance is the point where you have the least current flowing and subsequently where you least need power.
But when driving it with a current source, resonance represents the point where you most need power rather than where you least need power.
Yeah, but then that whole Zen notion of a direct connection between the amplifier's output and the speker's voice coil without a bunch of intervening passive components (which is what many find so attractive about single driver fullrange speakers in the first place) flies out the window and now things aren't nearly as simple as they were.
se
SY said:
Getting back to my original point, yes, that's where you most need voltage with respect to a current source, but that's because shoving the same amount of current through a higher impedance results in a greater voltage drop across it and unless Ohm's Law has been repealed, that's where you most need power, not where you least need power.
se
Steve Eddy said:
The point is that unless you are trying to get a peaked bass, you've got to, one way or another, reduce the voice coil current over that frequency range. Whether it's EQ or conjugate networks or velocity feedback (a la Mills/Hawksford), you end up putting less voltage across the driver near and at resonance. And Ohm's Law is satisfied.
Hello Yves, the actual circuits we are talking about are almost the same as a regular pentode mode amp, its more in how you think about it than in different circuits.
A straight pentode output stage is already a good transconductance amp, the OPT reflects the load impedance to the tube, which adjusts its output voltage to maintain the current set by the grid voltage. If you have a uniform impedance you also have a good voltage amplifier.
The fuss is about certain single driver speakers that have very high Qs, which give a falling bass response when driven by voltage amp. But when you put them on a straight pentode amp the transconductance causes greater voltage to be applied as the impedance goes up in the LF area, causing greater output and compensating for the LF rolloff. This will automatically happen when such a speaker is connected to a pentode connected in pentode mode.
The problem is that the designer has probably designed the output stage to drive 8 ohms, so it runs out of voltage swing with the high impedance. Thats why you use a lower turns ratio on the OPT so you have enough voltage swing to handle the higher impedance.
The circuits don't look particularly different, but the OPT parameters, operating point on the tube, the particular tube chosen etc will probably be different than for a "voltage amplifier'
BTW I gave the wrong tube type in my previous post, its the 6JH8 BDT tube.
John S.
A straight pentode output stage is already a good transconductance amp, the OPT reflects the load impedance to the tube, which adjusts its output voltage to maintain the current set by the grid voltage. If you have a uniform impedance you also have a good voltage amplifier.
The fuss is about certain single driver speakers that have very high Qs, which give a falling bass response when driven by voltage amp. But when you put them on a straight pentode amp the transconductance causes greater voltage to be applied as the impedance goes up in the LF area, causing greater output and compensating for the LF rolloff. This will automatically happen when such a speaker is connected to a pentode connected in pentode mode.
The problem is that the designer has probably designed the output stage to drive 8 ohms, so it runs out of voltage swing with the high impedance. Thats why you use a lower turns ratio on the OPT so you have enough voltage swing to handle the higher impedance.
The circuits don't look particularly different, but the OPT parameters, operating point on the tube, the particular tube chosen etc will probably be different than for a "voltage amplifier'
BTW I gave the wrong tube type in my previous post, its the 6JH8 BDT tube.
John S.
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.