Protecting your valuable output transformer
When using fixed-bias you don’t want to take any risks and damage your output transformer. What happens if your output valve (e.g. a transmitting power valve) makes false contact at the grid or the bias supply fails to start for whatever reason. Well, your valve will conduct fully and probably melting the anode and if not damaging the primary winding of your OT. Both things are catastrophic and you don’t want to try it…too expensive!
You may think there is a simple solution: a fuse. We all know fuses are inaccurate and can be dangerous if they don’t blow properly. Also HT fuses are expensive and they don’t come up with a wide range of values.
Rod Coleman pointed me out time ago to include a classic crowbar protection circuit using an SCR device to shunt the main HT supply and blow a fast fuse securely. This will be very quick and avoid any damage at all at the output stage.
The circuit is very simple to implement and to tweak to your own requirements. The sensing resistor has to be placed at the return loop of the cathode of the output valve. The sense resistor (R3 below) will develop a voltage which is proportional to the anode/cathode current. The R4/C1 RC stage provides a low pass stage to ensure the SCR is not triggered by any small variation in the current. Also there is a 12V zener connected between the SCR gate and cathode to protect it. When the SCR is triggered, then the R2 (a big 7W wirewound one) will shunt more than 500mA instantly and blow the fast fuse (which is not shown below) but is in series between the raw supply and R2.
I’m using an IXYS high-voltage thyristor which is the CS19 (800V/20A). Worst case scenario the shunt resistor (R2) will blow away and if you use on of the white ceramic ones, they will not catch fire.
My design is optimised to trigger at 200mA DC. The 4-65a is biased at 100mA and will peak at 170mA maximum. Is quite unlikely that a very low 20Hz tone will have that level as will probably kill my speakers but if that is the worst case scenario, then we can see that a transient signal of period 50ms has to be at least of 220mA peak to trigger the circuit:
This will clearly give sufficient headroom to protect the output stage from false triggers. If something goes wrong, the anode current will go up and when 200mA are reached, then the circuit will trigger and cut the HT supply by blowing a 500mA Fast Fuse. A fuse is a very decent price to pay in case of an accident.
Impact in sound is insignificant as there is no current flowing through the gate in normal operation and interaction with the RC capacitor is minimum.
Hi Ale,
How would you implement this protection when using filament bias for a DHT. For example using filament bias for a PSE 4P1L the resistor value Rk( R fil) is 15ohm. What value can you use for sensing in this case. Having expensive OT’s leads to the need of somekind of protection that would have the minimum impact on sound.
What if that overcurrent sensed would disconnect a relay used also for delay B+?
Best regards,
Radu
Hi Radu,
If you are using filament bias you have then a good protection and I can’t see why going through adding the complexity of this addition. The crowbar is intended for fixed bias.
If you want to implement this you should add the sensing circuit in the cathode return path to avoid the variance of the filament current impacting the performance of this circuit
Ale
Hi,
In my GM-70 SE I use a microcontroller to check the instantaneous and average current through each tube and cut the PSU transformer primary in case of biasing problems. It’s more expensive than the crowbar circuit (~15€), but does other tricks, too.
Hi Vlad,
Please, could you give me more info about your circuit?
Regards.
Hi Ale,
With this setup, when SCR is triggered, a maximum current of 600mA flow by the fuse. This is barely 1.2 times the nominal current value. Most of fuses will blow at 2,75 times the nominal current in 0,2 to 2 seconds. With 1,2 times the current, time of blowing can be of minutes. At this time, resistor can be fired.
I am not sure that 500mA fuse will be the righ value, probably a little low value works better.
Hi Luis
Good observations, I changed time ago the resistor and the fuse type. I currently have a 300R resistor and a microwave type HV fuse to avoid any issues. Having said that, I’m not looking forward to any further accidents to test again the crowbar circuit!
thanks, Ale
Hi Ale,
Another matter would be the RC filter with the values shown. With 47R and 100uF the constant time will be only 4.7 ms.
It seems more suitable to me that 100R and 470uF we get 47 ms of constant time, for a 21 Hz cut-off frequency.
Hi Luis, yes agreed. A good update for future. The amp will be put out back to storage in a couple of months given I have a baby girl so will bring in a safer amp to have around 🙁
Hi Ale,
I am designing a testing circuit for crowbar with these requirements:
1) Two PSU’s, Low voltage, a fixed voltage of 15V@1A and variable 0-15V@1A
2) Light indicator when SCR is on
3) Light indicator when Fuse is blown
With this setup I believe we will be sure when Fuse is blown and how much voltage is needed for triggering the SCR without hazardous high voltage around.
I will send you an e-mail with my setup.
Regards,
Luis.
Hi Ale,
My final setup is:
R = 220R
F = 500mA Slow Blow
HT = 425V, when SCR is triggered, current is 1.9A, for assuring blow fuse.
This afternoon, i leaves the RCAs input bad connected to the source, and with amp working i went to fix the bad connection. When i connected an RCA to amp, a big noise was created at the moment of connection was made. This was capable of triggering the SCR and fuse was blown.
My HP filter is R=47 and C=1000uF, I suppose that big noise is very low freqency and makes to shoot the SCR….
have you get the same experience?
Regards,
Luis.
Oh! my God.
Not the fuse was blown! was the series R with SCR that was blown!
The fuse is ok, because the R blowns before.
I am considering another protection based on Relay over primary of power xmer. Using Two Coil latching Relay and Reset by button seems more secure to me.
On the other hand, a fuse in signal path could have some drawbacks in THD, I don’t know if this high level of signal were affected by this.
Regards.
Hi Luis,
What is the rating of your 220R? I told you so about the fuse specs 🙂 You need a HV rated fuse for this. Did you change the fuse type as previous comment you mentioned that the fuse was blown with the low frequency noise?
A fuse in the supply line will not introduce any distortion as any variation on its resistance is insignificant compared to the low current and the supply output impedance in my view. A different story is to place them on the output before the speakers. I measured the response and is perfectly fine.
A tricky circuit to set it up correctly
When I went to replace fuse I saw blow R… and un-blow fuse.
R is 5w rated. Instantaneous power dissipation must be a bunch (820 watts). Heat transmission and thermal innertia would be very slow and very high temp melt the wire of resistor. Seeing to filament of fuse it seems intact. no melting or other signs.
I think the same as you about distortion by fuse in this position, but I’m goint to do some test about.
Never had this issue before, I plug in and out cables all the time. It seems to me you may have some ground loop or DC issue around and a stepped change may trigger the SCR. What about increasing the RC constant then?
Cheers
Ale
It is very common issue depending of RCA connector type you use.
If live connects *before* that GND pole, you get a loud bbrrrrr.
Female RCAs has live connections at different depths, some connect two poles at same time (or close) and another in a worst way.
and if you plug the RCA slowly… (as I did) matters go in the worst case possible 🙂
Sounds like a ground loop to me. A 1:1 input transformer would be great for galvanic isolation!