Home |
Search |
Today's Posts |
#1
|
|||
|
|||
µP-controlled bias
Worth thinking about? Ideas?
cheers, Ian |
#2
|
|||
|
|||
Ian Iveson wrote: Worth thinking about? Ideas? cheers, Ian What is µP?? Patrick Turner. |
#3
|
|||
|
|||
Ian Iveson wrote:
Worth thinking about? Ideas? cheers, Ian Not micro-processor but there are analogue bias servos in both Morgan Jones 'Valve Amplifiers' & Kevin O'Connor's 'Principles of Power' books. Both look promising enough to try sometime. The technique would be especially useful when dealing with tubes such as the 6080/6AS7 family where cathode bias resistors dissipating a lot of heat are normally used. Cheers, John Stewart |
#4
|
|||
|
|||
"Ian Iveson" wrote in message
... Worth thinking about? Ideas? cheers, Ian I think that's an interesting thought Ian, but wouldn't a microprocessor (forgot the ASCII code, show-off) generate a tremendous amount of heat with all them tubes needed, even for a 4 bit one? west |
#5
|
|||
|
|||
Mark Harriss wrote: Ian Iveson wrote: Worth thinking about? Ideas? cheers, Ian Doesn't VTL already do this on their new designs?. http://www.vtl.com/pages/Amplifiers/Siegfried/ Regards Mark Harriss Yes, it does look like VTL does use a vast array of silicon circuitry in tiny chips to try to get the eternal problem of biasing correct in what must be an expensive amp. I faced some bias dilemnas in the the 300 watters i made with a dozen 6550. These are class AB amps, and have simple cathode resistors, one for each tube, and each bypassed with 1,000 uF, so a total of 12,000 uF is used to bypass all those cathodes. Then i have a simple pair of power bjts which dynamically bypass excess nonlinear charge up currents in the bypass caps if tested with a sine wave up to full power. I posted the explanation of the principle about 2 years ago here and the schematic at abse. Nobody took much notice. But active grid voltage adjustment to control bias levels is not a very easy or effective way to control bias drifts. Simple cathode bias is great because its simple, and reliable, but all AB amps with CB have a problem when the amp is powered well into class AB and then the cathode bypass caps charge up, and stay charged until the signal level has dropped and during this time there may be 60v across the bypass resistor instead of the wanted 45v to get minimal thd. If there was some way to bypass the charge up current when it happened, the voltage across the cathode caps could be held very nearly constant, and large sudden transients in music would not allow the cathode bias to change. Testing with sine waves would be like testing with fixed grid bias. Anyway, that is what i have achieved by using a couple of cheap generic power bjts on a small heatsink, along with a few resistors and one cap and a few diodes. One R is a current sensing R, and if the voltage exceeds the base emitter voltage of a bjt, the bjt turns on and bypasses the caps, so the more dynamic Ik there is, the more the excess Ik is shunted through the power bjts, instead of charging up the caps, and so Ek is clamped, but the DC regulation of the bias by Rk is allowed to carry on unhindered. Thd is about 6 times less doing it the way I have compared to class AB and tested with a sine wave to clipping, and thd is virtually the same as fixed bias. Cost was about $20 to build my "dynamic bias regulator". Selling a client an amp system with 24 fixed bias pots to fiddle with would be a disaster. Having active protection against excess Ik in any one tube or many tubes is very easy to achieve with a few signal bjts, and R&C components which will be **far more** easily understandable and serviceable in 20 years than anything from VTL. There is just ***no need*** for microprocessors in vacuum tube amplifiers of any kind. All parts of tube engineering should be rugged, simple, effective, sound well and be easily understood and serviceable. Tube engineering that strays from these ideas becomes fragile, overly complex, impossible or expensive to service, shortlived, and a ****ing nuisance after awhile. Microprocessor bias control won't make an amp sound better, especially a high powered one, where the guy has bought it for the effortlessness of the power, the huge class A reserve, the limitless power ceiling, the low inherent distortions, and to impress all his mates. Usually such hi-fi amps have to make only maybe 5 watts average power levels, like any other amp with 50 watt ratings. Patrick Turner. |
#6
|
|||
|
|||
"Patrick Turner" wrote in message ... There is just ***no need*** for microprocessors in vacuum tube amplifiers of any kind. Much more of interest to me at least, is the fixed bias system with LED indicator (traffic light) which I believe I saw on one of your amps, Patrick. Can you outline the method of operation, please? Iain |
#7
|
|||
|
|||
Mark Harriss wrote: Patrick Turner wrote: Ian Iveson wrote: Worth thinking about? Ideas? cheers, Ian What is µP?? Patrick Turner. microprocessor, or microcontroller: usually a sub $10 computer on a chip. They can have inbuilt A-D and some form of D-A like pulse width modulation. They do have advantages over analogue circuits for some things, especially if you want to perform some complex calculation on whatever you are measuring. They are getting to the point where they are cheap: sub $1 for some types and have enough speed and power and memory to really be useful I just bought a US $10 Zilog developer kit which has an 8 pin microcontroller that could be used to measure something in 10 bit resolution and then act on it, say cathode bias and shut off the PSU if it exceeds parameters, it could also act as an hour meter and hook up to a PC via the com port to show a graph of tube bias settings over it's life. not much use for a small valve amp but handy for a transmitter or something. The only parameter of importance is the cathode current. If the DC bias current exceeds a certain value of say 50% above normal for 5 seconds, the amp is shut down. A simple circuit using discrete parts can protect 1 or 50 tubes against excess cathode current. However, one other parameter not much considered by anyone is to have a circuit which detects whether an amp is experiencing a load value less than the minimum recomended, say 2 ohms, even at low levels of signal, and then turns the B+ off. This means the output current needs to be compared to the output voltage, and if the comparison is applied to a simple differential pair and the balance of current to voltage means there must be a 2 ohm load, click, and the B+ shuts off. Speaker shorts, or drivers seizing in their magnetic gaps and then having their Z go low would all cause a load of below 2 ohms, and thus trigger the amp to shut down. All SS amps ***NEEED*** this protection, but I don't know of a single one with any such protection. Tubes survive occasions of excess Ik and can turn red with twice the Pda and still have a long life after that event. But my protection circuits will shut down an amp well before any anode glows red hot. Microprocessors are not needed in tube amps at all. Patrick Turner. Regards Mark Harriss |
#8
|
|||
|
|||
The problem with an MCU inside any amplifier is the potential for noise
generation -- RFI/EMI can cause non-linearity in high gain/low noise circuits. I've used Microchip and Atmel MCU's in other audio projects (as well as the BasicStamp when I started seven or eight years ago.) J "Ian Iveson" wrote in message ... Worth thinking about? Ideas? cheers, Ian |
#9
|
|||
|
|||
On Tue, 14 Jun 2005 06:38:00 +0000, Ian Iveson wrote:
Worth thinking about? Ideas? I considered this a while ago. First, the op-amp has a rather low voltage limit so my idea was to use an FET between the grid resistor & the -ve bias voltage (powering the op-amp between bias -ve and a small +ve supply). By monitoring the cathode volts of one p-p valve & comparing that with a fixed preset voltage it would set the bias point. A second op-amp would then compare the two cathode voltages & adjust the second output's valve current. It could be suitably slugged with capacitors to prevent silly distortion. One pot to set. All just an idea - I haven't tested it. -- Mick (no M$ software on here... :-) ) Web: http://www.nascom.info Web: http://projectedsound.tk |
#10
|
|||
|
|||
On Tue, 14 Jun 2005 06:38:00 +0000, Ian Iveson wrote:
Worth thinking about? Ideas? Of course, my op-amp idea isn't uP, but that has problems because of the resolution of the output. uPs arn't generally really nice analogue outputs (at my sort of prices anyway!). -- Mick (no M$ software on here... :-) ) Web: http://www.nascom.info Web: http://projectedsound.tk |
#11
|
|||
|
|||
Ian Iveson wrote:
Worth thinking about? Ideas? cheers, Ian Saw at tubecad a fairly simple circuit using an op-amp (not part of the audio path) to sense the cathode current and adjust the bias to get the current to a desired level. I don't think a uP would do it any better. |
#12
|
|||
|
|||
"John Stewart" wrote
Not micro-processor but there are analogue bias servos in both Morgan Jones 'Valve Amplifiers' & Kevin O'Connor's 'Principles of Power' books. Both look promising enough to try sometime. The technique would be especially useful when dealing with tubes such as the 6080/6AS7 family where cathode bias resistors dissipating a lot of heat are normally used. Right. I haven't seen O'Connor's book. Also on Norman Koren's beautiful site, http://www.normankoren.com/Audio/TENA.html Servos have problems, such as dealing with even harmonic distortion and class B operation. But the fundamental one is that they are continuous. The only thing a bias should interact with is the state of the valve, and that only needs checking and adjusting occasionally. So, for example, if you are using a µP for the remote switching, volume control and display in an integrated, it wouldn't take much extra processing to check the bias every time you switch it off (if it is up to full temperature or has been on for a minimum time), and reset according to that check every time you switch it on. Degradation should be slow enough that you could just increment by a fixed small voltage each time a valve passes an upper or lower threshold, like a simple thermostat. The processor can sleep and maybe the clock can be switched off for most of the time. The remote would wake it up just when processing is required. cheers, Ian |
#13
|
|||
|
|||
Iain M Churches wrote: "Patrick Turner" wrote in message ... There is just ***no need*** for microprocessors in vacuum tube amplifiers of any kind. Much more of interest to me at least, is the fixed bias system with LED indicator (traffic light) which I believe I saw on one of your amps, Patrick. Can you outline the method of operation, please? In all my PP amps I fit bias balance indicators where possible so that an owner can see from across the room if the bias drifts by more than 3 mA per output tube. Turning up the volume into a short, or serious clipping also makes the balance falter, so such conditions are easily spotted across the room. In an early re-do of an ST70 6 years ago I had a green led to indicate it was on, and all ok, then an orange if a slight problem occurred with bias balance, and then a red if the SCR and relay controlling the B+ had been tripped. The guy who bought this re-engineered ST70 did have a problem 18 mths later when an EL34 had a heater wire connection go dry, which made one channel single ended. He phoned to say the amp kept turning on its orange light when music played. I had the problem fixed while he waited. In the 300 watters, a balance monitor is unwieldy because there are 12 tubes to worry about, not just 2 or 4. So thus I used cathode bias, and the cathode DC voltage is monitored and if any cathode voltage goes above a certain value, an SCR fires and a relay turns on a red led and turns the B+ off. The same led also responds to clipping. Clipping will occur if a speaker becomes shorted, or goes to a very low value somewhere along the band, or someone uses too much volume. In the case of cathode detect circuits, let's say you have 4 cathodes. ( could be 40 cathodes though ). If you have fixed bias a 10 ohm current sensing Rk can be used, or if cathode bias, the current sensor is alrready there, its Rk. But consider the fixed bas amp. The signal voltage from the 10 ohm R is smoothed with a about 5k and 220uF cap. The output from this LPF is taken to the anode of a diode, 1N918 perhaps, and so for 4 cathodes you have 4 RC circuits, and 4 diodes. The cathodes of all 4 diodes can be taken to common point from where a voltage divider is used so that if any one or all of the 10 ohm voltage/s rise to 1v due to 100mA of tube current, then that exceeds the turn on forward voltage of the diode and a DC voltage will appear at the divider. This signal can be used to and turn on a two transistor differential pair to work balance leds and or work a sensitive gate SCR, C106D is what I use, and that works a relay. There are variuous ways to do all this, but basically I have a simple logic curcuit using LPF and diodes to enable protection of any number of tubes. I don't have the schematics of these circuits at my website because anyone who is a keen hobbyist can easily work such a scheme out for himself. They shouldn't need me to spoon feed them. Another good scheme with fixed bias is to have a two colour led so that yellow appears if bias is too low, red if too high, and both remain out if bias is just right, and again a handful of very cheap discrete parts can be used to achieve all this. Nobody should ever need a microfukkinprocessor. Patrick Turner. Iain |
#14
|
|||
|
|||
Ian Iveson wrote:
Worth thinking about? Ideas? cheers, Ian Doesn't VTL already do this on their new designs?. http://www.vtl.com/pages/Amplifiers/Siegfried/ Regards Mark Harriss |
#15
|
|||
|
|||
Patrick Turner wrote:
Ian Iveson wrote: Worth thinking about? Ideas? cheers, Ian What is µP?? Patrick Turner. microprocessor, or microcontroller: usually a sub $10 computer on a chip. They can have inbuilt A-D and some form of D-A like pulse width modulation. They do have advantages over analogue circuits for some things, especially if you want to perform some complex calculation on whatever you are measuring. They are getting to the point where they are cheap: sub $1 for some types and have enough speed and power and memory to really be useful I just bought a US $10 Zilog developer kit which has an 8 pin microcontroller that could be used to measure something in 10 bit resolution and then act on it, say cathode bias and shut off the PSU if it exceeds parameters, it could also act as an hour meter and hook up to a PC via the com port to show a graph of tube bias settings over it's life. not much use for a small valve amp but handy for a transmitter or something. Regards Mark Harriss |
#16
|
|||
|
|||
If you are just monitoring the heat--what about a strategically placed
Kolitron tube? "Ian Iveson" wrote in message ... Worth thinking about? Ideas? cheers, Ian |
#17
|
|||
|
|||
uP are really cool for all kinds of stuff, but my objection to them in audio
circuitry is that they cause lots of noise. I agree with Patrick that you can do just about anything you want with a simpler solution in a tube amp--especially if you don't need to leave the class A world. That said I did make a uP controlled slow turnon circuit for my Manley 440's when I had them. Turn both those babies on at the same time and it would blow a 30 Amp circuit instantaneously. DIdn't need to use a uP--just happened to have the parts sitting around doing nothing. In those days I was doing a lot of 8bit embedded programming. "Ian Iveson" wrote in message ... Worth thinking about? Ideas? cheers, Ian |
#18
|
|||
|
|||
tubesforall wrote:
That said I did make a uP controlled slow turnon circuit for my Manley 440's when I had them. Turn both those babies on at the same time and it would blow a 30 Amp circuit instantaneously. DIdn't need to use a uP--just happened to have the parts sitting around doing nothing. In those days I was doing a lot of 8bit embedded programming. Care to share any details? I'm just getting started with PICs and yeah a uP is overkill but pretty damn handy at the same time. One guy has an 8051 displaying stats like bias current and temp to an LCD. Adam |
#19
|
|||
|
|||
"Mark Harriss" wrote
Right. Bit over the top though. I would go with everything except the "full reports" and the "continually checks bias", although it doesn't actually do that...it adjusts between music passages, which seems error-prone to me and is too clever by half. Really it only needs to be part of the start-up and shut-down procedures. The only thing a display would be useful for would be to demand a revalve. Doesn't VTL already do this on their new designs?. http://www.vtl.com/pages/Amplifiers/Siegfried/ One beef (there are a few) I have with servo bias is that it will tend to run the valves at a lower bias voltage when they are cold at start-up, and then progressively reduce until equilibrium. The combination of the thermal delay and that of the filter used to dynamically sense the bias current makes it hard to avoid "hunting". I would prefer to switch on and then wait a while for the amp to warm up before I use it. The VTL avoids the dynamic adjustment, but adjusts between music passages. Er, why? Does it have a problem with DC stability? It means the processor is listening all the time, and that *is* excessive. OTOH, just looking at an AR stereo amp costing UKP4,500 (a cheek considering dollars are worth bugger all), that has a line of 8 bias test points and pots taking up half the back panel. For why? Convenience? Perhaps you get a voltmeter and a screwdriver thrown in for that price? They are not for tech-heads, we can safely assume, so either the hapless owner worries and fiddles constantly, or takes it to the dealer occasionally. The dealer will switch it on, and wait two minutes before mis-setting. If the test points were inside the chassis it would be gone for a month while they found someone qualified to open it up, but at least there would be some chance the job would be done right. Now Americans say they went to the moon, and they want a stack of proper money for an amp that can't adjust its own bias? It's not rocket science. cheers, Ian |
#20
|
|||
|
|||
"Mark Harriss" wrote
The simple solution would be to have the micro save the bias from last time the amp was running and use it as a starting point, that is once the amp has been turned on long enough to warm up. The micro would only check for a fault condition in that warmup period until the amp has warmed up for a set period. A simple trip circuit would be a more reliable way of dealing with a fault IMO They did make it to the moon and their computers did have a few hiccups along the way: A "divide by zero error" at the halfway point that required a reset and some octant star readings out the porthole to initialise the computer with their current readings and some kind of overflow error on the lunar lander: A mission controller was awarded a medal for telling the astronauts to keep punching the reset button on the way to the surface as the computer was built in such a way that it was able to pick up where it left off and keep doing it's job. They'd be half way to the sun by the time windose reset, and the blue screen with the warning about properly shutting down would really wind them up. The level of effective organisation of NASA is truly impressive in reaction to a crisis, or so it appears. Rather better than the military, generally. The Russians no less so, perhaps more when they were using dozens of mathematicians instead of a computer. Europe always argues until things crash then blames the French for their unpredictable and volatile rocket engines. Anyway, what's become of Mikkel Simonsen? He was our µP guru. Perhaps he has defected? cheers, Ian |
#21
|
|||
|
|||
Adam Stouffer wrote:
tubesforall wrote: That said I did make a uP controlled slow turnon circuit for my Manley 440's when I had them. Turn both those babies on at the same time and it would blow a 30 Amp circuit instantaneously. DIdn't need to use a uP--just happened to have the parts sitting around doing nothing. In those days I was doing a lot of 8bit embedded programming. Care to share any details? I'm just getting started with PICs and yeah a uP is overkill but pretty damn handy at the same time. One guy has an 8051 displaying stats like bias current and temp to an LCD. Adam Check out the Zilog Encore XP development kits: $10 US with an 8 pin package and an inbuilt temp sensor/UART/A-D convertor etc. etc. Regards Mark Harriss |
#22
|
|||
|
|||
Ian Iveson wrote:
"Mark Harriss" wrote Right. Bit over the top though. I would go with everything except the "full reports" and the "continually checks bias", although it doesn't actually do that...it adjusts between music passages, which seems error-prone to me and is too clever by half. Really it only needs to be part of the start-up and shut-down procedures. The only thing a display would be useful for would be to demand a revalve. Doesn't VTL already do this on their new designs?. http://www.vtl.com/pages/Amplifiers/Siegfried/ One beef (there are a few) I have with servo bias is that it will tend to run the valves at a lower bias voltage when they are cold at start-up, and then progressively reduce until equilibrium. The combination of the thermal delay and that of the filter used to dynamically sense the bias current makes it hard to avoid "hunting". I would prefer to switch on and then wait a while for the amp to warm up before I use it. The VTL avoids the dynamic adjustment, but adjusts between music passages. Er, why? Does it have a problem with DC stability? It means the processor is listening all the time, and that *is* excessive. OTOH, just looking at an AR stereo amp costing UKP4,500 (a cheek considering dollars are worth bugger all), that has a line of 8 bias test points and pots taking up half the back panel. For why? Convenience? Perhaps you get a voltmeter and a screwdriver thrown in for that price? They are not for tech-heads, we can safely assume, so either the hapless owner worries and fiddles constantly, or takes it to the dealer occasionally. The dealer will switch it on, and wait two minutes before mis-setting. If the test points were inside the chassis it would be gone for a month while they found someone qualified to open it up, but at least there would be some chance the job would be done right. Now Americans say they went to the moon, and they want a stack of proper money for an amp that can't adjust its own bias? It's not rocket science. cheers, Ian The simple solution would be to have the micro save the bias from last time the amp was running and use it as a starting point, that is once the amp has been turned on long enough to warm up. The micro would only check for a fault condition in that warmup period until the amp has warmed up for a set period. They did make it to the moon and their computers did have a few hiccups along the way: A "divide by zero error" at the halfway point that required a reset and some octant star readings out the porthole to initialise the computer with their current readings and some kind of overflow error on the lunar lander: A mission controller was awarded a medal for telling the astronauts to keep punching the reset button on the way to the surface as the computer was built in such a way that it was able to pick up where it left off and keep doing it's job. Regards Mark Harriss |
Reply |
Thread Tools | |
Display Modes | |
|
|
Similar Threads | ||||
Thread | Forum | |||
KISS 121 by Andre Jute | Vacuum Tubes | |||
KISS 113 by Andre Jute | Vacuum Tubes | |||
Disabling M-Audio Transit USB Analog Input DC Bias | Pro Audio | |||
Improved AM Detector | Vacuum Tubes | |||
Toroidal OT question. | Vacuum Tubes |