I am home on leave so had some time to putter away in LTspice (dangerous). Looking at possible 6EM7 amps I decided to take a look at fixed bias. I normally think in terms of cathode bias for simplicity and lack of maintenance but I thought I should look at fixed in this case since as a low powered amp the likelihood of being driven into blocking is higher. Fixed bias allows me to direct couple a CF to the grid of the power tube eliminating that cap and also the cathode bypass cap. As a bonus I guess you get the ability to drive the grid slightly positive but that is really not a big concern for me.
Well the simulations were really encouraging and as a bonus it seems there is a non-trivial increase in max power output. If you are starting with relatively high power the extra isn't a huge deal but when you are starting at just a couple of watts it is kind of nice.
This is what I was looking at.
The FB loop is primarily to make for consistent gain since I intend to use it later as a phase for an input transformer split PP amp. I could use the 6EM7 small signal triode for gain and another higher current triode for the CF but that complicates the layout and wiring to keep gain stage away from PS and output. I have reduced C1 to .033uf as it seemed unnecessarily large.
I am curious as to whether the biasing scheme is reasonable. It would be fed by either a separate transformer tap or possibly its own transformer. R6 and R15 are the bias pot and the two 1 Meg resistors R16 and 17 are from the wiper to each end to prevent runaway if the pot's wiper goes open.
Well the simulations were really encouraging and as a bonus it seems there is a non-trivial increase in max power output. If you are starting with relatively high power the extra isn't a huge deal but when you are starting at just a couple of watts it is kind of nice.
This is what I was looking at.
The FB loop is primarily to make for consistent gain since I intend to use it later as a phase for an input transformer split PP amp. I could use the 6EM7 small signal triode for gain and another higher current triode for the CF but that complicates the layout and wiring to keep gain stage away from PS and output. I have reduced C1 to .033uf as it seemed unnecessarily large.
I am curious as to whether the biasing scheme is reasonable. It would be fed by either a separate transformer tap or possibly its own transformer. R6 and R15 are the bias pot and the two 1 Meg resistors R16 and 17 are from the wiper to each end to prevent runaway if the pot's wiper goes open.
Attachments
1. Just be sure that the quiescent conditions of the output tube do not cause it to go into
thermal run-away. There is no auto current protection that self bias has.
2. Be very sure that V3 bias supply is at full voltage . . . before the output tube U3 warms up, and U3's B+ comes on.
The smoke is so very hard to stuff back into the amplifier's parts.
3. Suppose that a small amount of grid current (class A2) gives 1 dB more output power.
Then, amplifier #1, a 1 Watt amplifier becomes a 1.25 Watt amplifier.
And, amplifier #2, a 100 Watt amplifier becomes a 125 Watt amplifier.
Suppose amplifier #1 drives a 102dB/Watt speaker
And suppose amplifier #2 drives an 82dB/Watt speaker
The two systems will have exactly the same maximum volume.
Conclusion: a 1 dB change in output power will also sound exactly the same, whether it is a low power amplifier, or a high power amplifier.
If the increase due to going from Class A1 to Class A2 is 3dB, those two amplifier and speaker combinations will have the same results in the sound level of both; if you compare the 3dB difference in A1 and A2 sound levels.
thermal run-away. There is no auto current protection that self bias has.
2. Be very sure that V3 bias supply is at full voltage . . . before the output tube U3 warms up, and U3's B+ comes on.
The smoke is so very hard to stuff back into the amplifier's parts.
3. Suppose that a small amount of grid current (class A2) gives 1 dB more output power.
Then, amplifier #1, a 1 Watt amplifier becomes a 1.25 Watt amplifier.
And, amplifier #2, a 100 Watt amplifier becomes a 125 Watt amplifier.
Suppose amplifier #1 drives a 102dB/Watt speaker
And suppose amplifier #2 drives an 82dB/Watt speaker
The two systems will have exactly the same maximum volume.
Conclusion: a 1 dB change in output power will also sound exactly the same, whether it is a low power amplifier, or a high power amplifier.
If the increase due to going from Class A1 to Class A2 is 3dB, those two amplifier and speaker combinations will have the same results in the sound level of both; if you compare the 3dB difference in A1 and A2 sound levels.
Last edited:
All good points thanks. Yeah, the extra power is just a little bonus. The main attraction to me is the promise of better overload recovery. Any suggestions on simple ways to be sure that the grid supply comes up first? Would we get some benefit from the large capacitance that the B+ will need to charge? I suppose an easy way is to use tube regulator for B+ and SS diodes for grid supply. Of course the tubes won't conduct until the heaters warm up while a SS PS for the grids should be almost immediate. Can one fuse the cathodes without messing up the sound in case of grid supply failure?
I've always liked this idea of Wavebourn's. Not sure how easily it could be adapted to your situation.
Link 2 is to a reposting and the post immediately below says more about its functioning.
He's posted this idea a few times if you want to search for more details.
Another idea.
There's also a bias circuit idea for hard full neg bias at turn on and then bringing it up to set value after time to let the tubes warm. Rod Coleman has a board for this .
HTH
Link 2 is to a reposting and the post immediately below says more about its functioning.
He's posted this idea a few times if you want to search for more details.
Another idea.
There's also a bias circuit idea for hard full neg bias at turn on and then bringing it up to set value after time to let the tubes warm. Rod Coleman has a board for this .
HTH
Last edited:
I kind of like the idea of using the cathode sense resistor as a "fuse" of last resort. Is there a rule of thumb on how much overload it takes to pop a resistor?
I think on run of the mill film resistors the behavior is variable and therefore not guaranteed reliable. The places I've seen engineers using them as fuses were with sand devices capable of dumping lots of current in a short time.
Fusible resistors listed at Digikey are 1 watt at the lowest and looking at Bourns' chart they would be too slow.
Looking at smaller power ratings with "defined interruption behavior" they are still not guaranteed to fail completely open. Vishay only guarantees rise of resistance to at least 10x value.
One Vishay datasheet for .3 and .5 watt resistors shows earliest failure time data starting at 3 Watt dissipation.
What about a fuse under the resistor? Or a relay in the cathode that would interrupt mains power if activated ?
Fusible resistors listed at Digikey are 1 watt at the lowest and looking at Bourns' chart they would be too slow.
Looking at smaller power ratings with "defined interruption behavior" they are still not guaranteed to fail completely open. Vishay only guarantees rise of resistance to at least 10x value.
One Vishay datasheet for .3 and .5 watt resistors shows earliest failure time data starting at 3 Watt dissipation.
What about a fuse under the resistor? Or a relay in the cathode that would interrupt mains power if activated ?
Attachments
Some of my amp designs use silicon for the negative supply and a tube rectifier for main B+. This allows grid bias to be either too negative, or at the proper value before the output tubes are warmed up.
In the Tubelab TSE amp the negative supply is SS. The B+ rectifier is usually a 5AR4. The output tubes are DHT's. Here the negative supply is at full voltage in milliseconds, and the DHT's warm up next. The tube rectifier is slower than the DHT's. The grid drive on the output tubes is a mosfet source follower in a circuit similar to your cathode follower. The bias is excessively negative until the main B+ comes up since it is part of the bias voltage divider. This works well even when a 5U4 or other DH rectifier is used. Some RC time constant juggling may be needed if both supplies are SS and fast warm-up output tubes are used (DHT's).
Issues can arise when silicon is used for both power supplies and a cathode follower is directly coupled to the output tube, since bias stability is dependent on who wins the warm-up race, the output tube or the cathode follower. Some tube dependent bias drift can be expected. Using a IDH cathode follower with a DH output tube is not always a good idea unless the circuit is designed such that the bias is always pulled to the negative rail until the driver warms up.
Fusing the cathode works fine in loss of bias situations. It is often less than successful in gassy tube runaway situations. A 1/4 amp or so "pico fuse" may save the tube or OPT in a bias loss situation.
I find that these Dale 1 ohm 2 watt resistors in the cathode lead blow like a fuse in extreme cases of "too much power." In this case I was finding the limit of some old 6L6GA tubes in AB2 on 500 volts into a 3300 ohm OPT. In both cases one of the cathode resistors blew due to a tube arc somewhere beyond 110 watts output. In both cases the tubes survived and still work today at a more sane power level.
They have prevented disasters in sweep tube amps where the tube can pass amps of current and my old HP bench power supply goes to 650 volts and has 1000 uF of output capacitance. An unfused screw up resulted in one of the OPT's becoming a magnet. It was a $16 test OPT, and it was successfully degaussed, but it still doesn't sound right and produces more THD than it's brothers.
I use a relay interlock for safety in my amps.
The -110V (or -200V on the larger amps) supply powers a relay that turns on the main B+ supply.
If the bias supply goes missing/fails, the main power is cut, saving the tubes. It also ensures the bias voltage is present on the grids before main power is there.
The -110V (or -200V on the larger amps) supply powers a relay that turns on the main B+ supply.
If the bias supply goes missing/fails, the main power is cut, saving the tubes. It also ensures the bias voltage is present on the grids before main power is there.
Thanks George. I will check out pico fuses. Also good thought about the driver v.s. output tube race. In this case the follower and the output tube are in the same bottle so the heater characteristics are the same so probably will be a photo finish. 🙂
Lots of those little buggers to choose from. I didn't even know they existed.
PICO(R) Fuses - Axial Radial Thru Hole PICO(R) Fuses - Littelfuse
Thank you also Kodabmx. That sounds like a very clever and simple solution. So as I understand it the normally open relay is powered by the bias supply and when energized connects the B+. Do you switch the Unfiltered side of the B+ or the filtered side?? If I use a separate tranny for the bias supply I could switch the primary side of the main tranny which has the advantage of switching a smaller AC signal but it would also mean a gradual ramp down if the bias supply fails which could lead to a significant over current time. Switching the heater supply is another option but again has a turn off delay.
Interesting topic.
Lots of those little buggers to choose from. I didn't even know they existed.
PICO(R) Fuses - Axial Radial Thru Hole PICO(R) Fuses - Littelfuse
Thank you also Kodabmx. That sounds like a very clever and simple solution. So as I understand it the normally open relay is powered by the bias supply and when energized connects the B+. Do you switch the Unfiltered side of the B+ or the filtered side?? If I use a separate tranny for the bias supply I could switch the primary side of the main tranny which has the advantage of switching a smaller AC signal but it would also mean a gradual ramp down if the bias supply fails which could lead to a significant over current time. Switching the heater supply is another option but again has a turn off delay.
Interesting topic.
Last edited:
Yes, NO relay with a 110VDC coil powered either directly from the -110V bias or from the -200V bias through a 10k resistor.
I switch the primary of the B+ transformer, but I have switched the B+ after the caps as well.
I switch the primary of the B+ transformer, but I have switched the B+ after the caps as well.
The really smart guys here tend to design circuits that are complex because they can. And, sometimes, complexity might be necessary. I get it.
I'm just a dolt, so I look for simple solutions that even I can (somewhat) understand.
According to the data sheet, the input section of a 6EM7 is typically biased at -3v. So, if you want to use fixed bias, why not just use a pair of AA batteries on the grid?
Are you open to using a different tube than the 6EM7?
Check out the 6N6G, which is also a dual, dissimilar, triode. It is already internally configured as a cathode follower directly coupled to the grid of the output section. It's also internally biased, just connect the "cathode" pin to ground. Power output is ~4w, which is probably about 2x that of the 6EM7.
It's also very easy to drive. I've been working on an SET using it and being driven by a 26, although I've had good results with a variety of low mu DHTs as input tubes. I'm using fixed (non-adjustable) battery bias on the 26s. A VR tube supplies the 26s and they're choke loaded using cheap Hammond 156Cs.
Obviously, if you want to design a complex circuit, this won't work and I apologize for the intrusion. I don't mean to spoil your fun or derail your thread.
A pair of AA batteries is OK. Better might really be the 'no leaky' big button cells. They're usually 3 volts and last for years and years without doing bad things.
There seriously is nothing wrong with using batteries. For 'awhile' designers were putting them in series with the grid! But, in that mode, they tended to be antennae, so not so good. Just connecting to ground, and thru a 100 kohm resistor to grid the 'normal' way is better.
100 kohm? Anything between whaterver-you-want-to-support as the input impedance of the stage, and the reality that as the impedance increases, flicker noise can become more significant.
GoatGuy
There seriously is nothing wrong with using batteries. For 'awhile' designers were putting them in series with the grid! But, in that mode, they tended to be antennae, so not so good. Just connecting to ground, and thru a 100 kohm resistor to grid the 'normal' way is better.
100 kohm? Anything between whaterver-you-want-to-support as the input impedance of the stage, and the reality that as the impedance increases, flicker noise can become more significant.
GoatGuy
I've got them going to ground at the moment, just as you've described.
Of course, I've also read recommendations that they be used in series and I was thinking of trying that. I'm wondering how much of an issue the "antennae" thing is, in practice.
I'm unfamiliar with "flicker noise". I have no way of measuring these things other than my ears but my current setup sounds nice.
I was under the impression that battery life, using either method of battery bias on the grid, was essentially the shelf life of the battery. That should be roughly 10 years for lithium and maybe 2 or 3 for alkaline. I've actually read that zinc carbon never need replacement but I'm not sure why.
Why would batteries in series with the grid last longer than a grounded setup?
One claim about series use confuses me. Using the grounded method requires the use of DC blocking caps on the input. I've read claims that when the battery is in series with the grid that the DC blocking caps are not needed. I read this on the TubeCad site, although no explanation was given.
Is this correct?
I've actually read conflicting claims about this and I don't understand why the blocking caps wouldn't still be necessary. If they can, in fact, be eliminated then that would seem to be more of an advantage than longer battery life, unless battery life was considerably shorter.
The battery can go in either place with about the same results. There must be a return path from the grid to the cathode with the battery and a resistor in it.
The battery can be connected directly to the grid with the resistor from the input of the amp to ground. No blocking cap is needed under most circumstances since the input is at ground potential. All tubes can draw small amounts of grid current if the grid gets near zero volts. This can force a current through the input resistor or the input source itself, especially if the input resistor is a large value. Some sources may not like this. This also puts a piece of ungrounded metal directly on the grid. In a high gain amp used near AM radio stations, or some cell phones (GSM is worse case) the battery can be a receive antenna.
Putting the battery to ground and a resistor to the grid puts a negative voltage on the grid, thus requiring a blocking cap.
If this is a guitar amp and you are driving it hard with a pedal board set on KILL, the bias will be seriously upset turning the grid into a rectifier which can be seen at the amp's input in the no cap approach. Either way if you anticipate a large driving signal, battery bias may not be your best choice.
I used to put one to three NiCad cells in series with the cathode for bias. The cathode current keeps them charged, forever. But as we saw with the 80's and 90's computer motherboards most NiCads will eventually leak making a corrosive mess. Today an LED in series with the cathode can do the same thing, you just need to find the right LED.
Compared to some of my more recent stuff this is a pretty simple circuit, just a typical two stage amp with a cathode follower added to kill blocking distortion. I tend to prefer mosfet followers, but everyone has their preferences.
The OP stated that he wanted to be able to use this as a driver someday.
Back in 2007 I needed a simple driver capable of producing a large voltage output for driving a cathode follower output stage. The 6EM7 fit the bill perfectly and was used in the design. I had a pretty large collection of 6EM7's and I had to try a few to find some with the same gain since this is a zero feedback design. There is a considerable variability in the Mu of the small triode from tube to tube, even of the same brand.
This was a really complex overall design with a DSP chip controlling the voltages on a vacuum tube amplifier. The stated purpose was to improve the efficiency of a class A tube amp by adjusting the operating conditions on the fly in a manner seen in class H solid state amps. It was an offshoot from work I was doing at Motorola on RF power amps used in cell tower transmitters.
The real purpose was to take a shot at winning a design contest sponsored by Microchip which ran in Circuit Cellar, a magazine that targeted embedded computer systems builders. I did not win the grand prize, but did win the "best use of SMPS resources" prize, which had a cash payout. This led to a published article in the magazine and another cash payout.
The article is now about 12 years old and vanished from their web site about 6 years ago so I am including it here. The amp schematic is in the article.
The battery can be connected directly to the grid with the resistor from the input of the amp to ground. No blocking cap is needed under most circumstances since the input is at ground potential. All tubes can draw small amounts of grid current if the grid gets near zero volts. This can force a current through the input resistor or the input source itself, especially if the input resistor is a large value. Some sources may not like this. This also puts a piece of ungrounded metal directly on the grid. In a high gain amp used near AM radio stations, or some cell phones (GSM is worse case) the battery can be a receive antenna.
Putting the battery to ground and a resistor to the grid puts a negative voltage on the grid, thus requiring a blocking cap.
If this is a guitar amp and you are driving it hard with a pedal board set on KILL, the bias will be seriously upset turning the grid into a rectifier which can be seen at the amp's input in the no cap approach. Either way if you anticipate a large driving signal, battery bias may not be your best choice.
I used to put one to three NiCad cells in series with the cathode for bias. The cathode current keeps them charged, forever. But as we saw with the 80's and 90's computer motherboards most NiCads will eventually leak making a corrosive mess. Today an LED in series with the cathode can do the same thing, you just need to find the right LED.
Compared to some of my more recent stuff this is a pretty simple circuit, just a typical two stage amp with a cathode follower added to kill blocking distortion. I tend to prefer mosfet followers, but everyone has their preferences.
The OP stated that he wanted to be able to use this as a driver someday.
Back in 2007 I needed a simple driver capable of producing a large voltage output for driving a cathode follower output stage. The 6EM7 fit the bill perfectly and was used in the design. I had a pretty large collection of 6EM7's and I had to try a few to find some with the same gain since this is a zero feedback design. There is a considerable variability in the Mu of the small triode from tube to tube, even of the same brand.
This was a really complex overall design with a DSP chip controlling the voltages on a vacuum tube amplifier. The stated purpose was to improve the efficiency of a class A tube amp by adjusting the operating conditions on the fly in a manner seen in class H solid state amps. It was an offshoot from work I was doing at Motorola on RF power amps used in cell tower transmitters.
The real purpose was to take a shot at winning a design contest sponsored by Microchip which ran in Circuit Cellar, a magazine that targeted embedded computer systems builders. I did not win the grand prize, but did win the "best use of SMPS resources" prize, which had a cash payout. This led to a published article in the magazine and another cash payout.
The article is now about 12 years old and vanished from their web site about 6 years ago so I am including it here. The amp schematic is in the article.
I used batteries in series with the grid stopper to the grid.
batteries are no worse antenna than a coupling caps of the same dimensions are.
Unless you draw grid current, the batteries work as long as their off-the-shelf life.
If you draw grid current, you are charging the battery.
I used batteries for negative grid bias from 1.5V to over 80V.
They all worked for me.
batteries are no worse antenna than a coupling caps of the same dimensions are.
Unless you draw grid current, the batteries work as long as their off-the-shelf life.
If you draw grid current, you are charging the battery.
I used batteries for negative grid bias from 1.5V to over 80V.
They all worked for me.
Last edited:
Great discussion guys. Charlie, I am pretty well set on the EM7s as I have several and have heard that they sound very nice. That 6N6G sounds pretty interesting though. My eventual plan is to use some Cinemag input trannies that have a split secondary to do phase inversion at the input with each side feeding one of these which of course then drive a PP OPT.
The small signal triode in the EM7 is know for being a bit variable in the parameters I decided to use it as the follower where that doesn't much matter and then use some plate to cathode feedback to a separate VAS so that the gain will be steady and equal for each phase. A little lower impedance driving the OPT isn't a bad thing either. Of course in this setup I am actually setting the bias for the output tubes by setting the grid voltage on the CF. That way I can easily experiment with operating points possibly even trying Class A and AB in the eventual PP form. I could use a series of 9V batteries to power the bias voltage divider but that would probably cost as much as building a normal grid supply given the stuff in my junk boxes.
In the mean time I have a couple EDCOR 5K SE OPTs so I can get the circuits working and hopefully debugged.
And then there is a preamp... 😀
The small signal triode in the EM7 is know for being a bit variable in the parameters I decided to use it as the follower where that doesn't much matter and then use some plate to cathode feedback to a separate VAS so that the gain will be steady and equal for each phase. A little lower impedance driving the OPT isn't a bad thing either. Of course in this setup I am actually setting the bias for the output tubes by setting the grid voltage on the CF. That way I can easily experiment with operating points possibly even trying Class A and AB in the eventual PP form. I could use a series of 9V batteries to power the bias voltage divider but that would probably cost as much as building a normal grid supply given the stuff in my junk boxes.
In the mean time I have a couple EDCOR 5K SE OPTs so I can get the circuits working and hopefully debugged.
And then there is a preamp... 😀
One point about the so called cathode "fuse" on the OP stage; in my experience most resistors go high not OC meaning the OP valve is biased very hot. All resistors I've measured that have gone high measure roughly 1m sending Vgk to about +30v. That said, such a resistor also limits Ik.
Liked your 1m R's across what I assume to be a pot in the -V bias. A good simple way for a non bias voltage situation is to use a relay powered by the bias V to connect HT/B+ to ground. No bias V, no HT. Also simple to use an RC time delay to energise the relay coil,hence bias protection & delayed bias in one.
Andy.
Liked your 1m R's across what I assume to be a pot in the -V bias. A good simple way for a non bias voltage situation is to use a relay powered by the bias V to connect HT/B+ to ground. No bias V, no HT. Also simple to use an RC time delay to energise the relay coil,hence bias protection & delayed bias in one.
Andy.