I am finalising the design of a large parallel push pull amplifier that has occupied much of my spare time over the last few years. Having finally gotten to the kind of performance I was looking for, the last piece of the puzzle is protection circuitry.
I understand clearly that the other measures I am putting in place (basically fuses, thermal protection on the regulated PSU and current sensing on the cathode resistors of the output tubes) will not save me versus an open circuit, because the problem is flyback voltage generation not current.
What I am contemplating is this - bolted to the case and across the OPT secondary a large triac (say 20A) with a (say) 50V bidirectional tranzorb diode to the gate (this is a 120W or so output amp into 8R).
Output goes open circuit, voltage on OPT secondary goes north, tranzorb conducts, triac fires, energy gets dumped into case as heat and we all go home happy.
Seems fool proof - is there a flaw that the experienced guys can see in my scheme!?
I understand clearly that the other measures I am putting in place (basically fuses, thermal protection on the regulated PSU and current sensing on the cathode resistors of the output tubes) will not save me versus an open circuit, because the problem is flyback voltage generation not current.
What I am contemplating is this - bolted to the case and across the OPT secondary a large triac (say 20A) with a (say) 50V bidirectional tranzorb diode to the gate (this is a 120W or so output amp into 8R).
Output goes open circuit, voltage on OPT secondary goes north, tranzorb conducts, triac fires, energy gets dumped into case as heat and we all go home happy.
Seems fool proof - is there a flaw that the experienced guys can see in my scheme!?
Sort of. The protection needs to primarily (pun intended) focus on the primary winding of the OT - as that is where the main risk of damage is likely imho.
There are a number of good forum threads on the topic, and many forms of protection used over the decades.
I even put an brief together on the topic for another website.
http://dalmura.com.au/projects/Output transformer protection.pdf
There are a number of good forum threads on the topic, and many forms of protection used over the decades.
I even put an brief together on the topic for another website.
http://dalmura.com.au/projects/Output transformer protection.pdf
Thanks for the link to your paper, will try to go over it this weekend, lots of info there... In the meantime, do you have any suggestion on making a safe amp head switcher like the Radial's Headbone VT?
The idea with triac is not good at all, in case of over-current you need to disconnect B+.
This can be done simple way (let's say with circuit breaker), or with Hall effect current sensor wired to Arduino with relay(s) or triacs, which can be connected to B+ of each channel, or even to each output tube.
As an alternative, you can build current monitoring system with Arduino and ballast resistor, follow link attached.
Measuring Current with the Arduino
This can be done simple way (let's say with circuit breaker), or with Hall effect current sensor wired to Arduino with relay(s) or triacs, which can be connected to B+ of each channel, or even to each output tube.
As an alternative, you can build current monitoring system with Arduino and ballast resistor, follow link attached.
Measuring Current with the Arduino
One obligatory addition is to put a resistor across the OT secondary (e.g 470R, 5W). This at least stops the secondary load from going infinite if the speaker is disconnected, so provides a modicum of damping for the primary, at very little cost. If there is already a Zoble filter then this will help too, of course.
You can also put a rever-biased diode/s from each power valve anode to ground. This clamps any possible negative-going flyback voltages.
I also have pondered the use of gas-tube spark supressors across the primary. These can be bought with specific break-over voltages and do not have the soft turn-on problems of MOVs.
You can also put a rever-biased diode/s from each power valve anode to ground. This clamps any possible negative-going flyback voltages.
I also have pondered the use of gas-tube spark supressors across the primary. These can be bought with specific break-over voltages and do not have the soft turn-on problems of MOVs.
Last edited:
Merlin, that is a neat device to think about placing across each half-primary, or from anode to anode. If both tubes in a PP config were in cut-off and an arc was triggered then that would be one scenario. Another is if one tube was conducting. And there may not be enough energy in everyday common leakage inductance induced flying voltage to start an arc, although it may provide some suppression from the device gas starting to glow. Testing what happens would make for an interesting rainy day.
Last edited:
The problem I see with the arduino is that the energy is already stored in the transformer. By the time the current is sensed, the field is going to collapse and an arch form.
The resistor, snubber, reverse biased diodes seems to be a more effective and reliable approach. Passive devices have reaction times orders of magnitude faster than a microprocessor.
The resistor, snubber, reverse biased diodes seems to be a more effective and reliable approach. Passive devices have reaction times orders of magnitude faster than a microprocessor.
I think LinksGuru misunderstood what this thread was about- OT protection, not power supply overcurrent protection.
To a man with an arduino, every problem is a nail. That needs arduino.
Hmmm
I do have overcurrent protection on the output tubes - basically 10R resistors are used from the KT88 cathodes to ground. I send this into a low pass filter with a TVS diode across it to protect the op amp comparator which then when a reference voltage is exceeded disconnects the main power transformer in the amplifier by switching out both poles of the primary.
I had thought of a resistor of 330-470R across the output, but I have seen this not work disastrously in guitar amps when leads get pulled at very high outputs so I wanted something better.
I guess the problem with sensing anything from the secondary is that by the time that happens there is already a lot of energy stored in the transformer, and the primary (500V B+ in the case of my design) could be heading towards the maximum voltage (1kV for the Lundahl LL1693).
I guess I'll stick to the resistor and the reverse-biased diodes! I work with computers and I could not face fighting with an Arduino to do anything!🙂
I do have overcurrent protection on the output tubes - basically 10R resistors are used from the KT88 cathodes to ground. I send this into a low pass filter with a TVS diode across it to protect the op amp comparator which then when a reference voltage is exceeded disconnects the main power transformer in the amplifier by switching out both poles of the primary.
I had thought of a resistor of 330-470R across the output, but I have seen this not work disastrously in guitar amps when leads get pulled at very high outputs so I wanted something better.
I guess the problem with sensing anything from the secondary is that by the time that happens there is already a lot of energy stored in the transformer, and the primary (500V B+ in the case of my design) could be heading towards the maximum voltage (1kV for the Lundahl LL1693).
I guess I'll stick to the resistor and the reverse-biased diodes! I work with computers and I could not face fighting with an Arduino to do anything!🙂
Hi,
I agreed with LinuksGuru using a micro to read the cathode current and if the current is it lost the micro will immediately enable a triac with dummy resistors connected from plate to ground to discharge the high voltage. The resistors need to be sizes depending of the current so not to burn the OT. I will use the ACS712 Hall effect current sensor to do the monitoring for the cathodes current since both are tied together to ground. It is a simple circuit and a good idea. Attached it is the schematic showing on how to do it.
I agreed with LinuksGuru using a micro to read the cathode current and if the current is it lost the micro will immediately enable a triac with dummy resistors connected from plate to ground to discharge the high voltage. The resistors need to be sizes depending of the current so not to burn the OT. I will use the ACS712 Hall effect current sensor to do the monitoring for the cathodes current since both are tied together to ground. It is a simple circuit and a good idea. Attached it is the schematic showing on how to do it.
You're right, it won't. I experimented with this years ago, deliberately driving a tube amp into an open circuit. I found that the 220 or 470 ohm resistor across the OT secondary did a good job of protecting the OT from spikes, but I also found that the screens started to overheat badly before the level of drive got high enough to produce spikes. So, spike protection and screen current limiting complement each other.
I have seen old schematics that showed a small air spark gap across each half-primary.
I have seen old schematics that showed a small air spark gap across each half-primary.
I do not know if this my help but do not forget that the micro can turn off the incoming AC also. The micro can do this in micro seconds.
More food for thought
I have just very cautiously tested the amp with no load. It is unstable oscillating at 1.5MHz. Glad I did that! Bode plotting time, and realised something I had taken out of the circuit had left me with too much HF gain / wrong values in the feedback compensation loop. Running the amp with no output load had exposed feedback loop instability. Order has now been restored and the amp is stable with no load and no input.
As I applied an input with no load the anode voltage begins to climb rapidly (although there is no impact on the output waveform - a square wave still looks very square) , and as soon as I got to the point where 1/2 of the output tubes cut off I begin to get flyback spikes. They are not that big because I did not want to go much further with this experiment in terms of increasing the input, and I am using a fully regulated B+ and the output of the supply has a reverse biased diode already there. However that is at the far end of the OPT primary so I have added a pair of UF5407 reversed across the output tubes, along with 330R across the output secondary.
That improves things a bit (much less anode voltage spiking), but I suspect full output open circuit would be smokeroo time.
So I will keep the diodes and output resistor in place as may fallback plan, order the triac and TVS diodes and experiment!
I have just very cautiously tested the amp with no load. It is unstable oscillating at 1.5MHz. Glad I did that! Bode plotting time, and realised something I had taken out of the circuit had left me with too much HF gain / wrong values in the feedback compensation loop. Running the amp with no output load had exposed feedback loop instability. Order has now been restored and the amp is stable with no load and no input.
As I applied an input with no load the anode voltage begins to climb rapidly (although there is no impact on the output waveform - a square wave still looks very square) , and as soon as I got to the point where 1/2 of the output tubes cut off I begin to get flyback spikes. They are not that big because I did not want to go much further with this experiment in terms of increasing the input, and I am using a fully regulated B+ and the output of the supply has a reverse biased diode already there. However that is at the far end of the OPT primary so I have added a pair of UF5407 reversed across the output tubes, along with 330R across the output secondary.
That improves things a bit (much less anode voltage spiking), but I suspect full output open circuit would be smokeroo time.
So I will keep the diodes and output resistor in place as may fallback plan, order the triac and TVS diodes and experiment!
- Status
- Not open for further replies.