Using Solid State Rectifiers To Replace Valves - Inrush Damage?

[Bazz4CQJ]

"Chunky" series resistors will sit there forever giving off heat. A resistor/zener combination across the HT will only pass current for ~10-15 seconds at switch on.

B

[Gabe001]

I have just finished a mullard 3-3 stereo and used solid state rectification. It required a 200ohm (25w) surge resistor to bring the voltage down to spec, with a 280-0-280 secondary. I've never had problems with solid state rectifiers, but if you are concerned about surge voltage and inrush current, you can always put a thermistor in the transformer primary for peace of mind.

Also good practice to put a fuse in the centre tap, in case the diode shorts.

A couple of pictures of mine in case it helps. You can see the surge resistor bolted to the aluminium plate and the 2 fuses (HT and LT)

[nemo_07]

Originally Posted by E93AFAN

... yes I was intending to calculate the value of a chunky resistor (or two) to reduce the voltage to that expected from the EZ81, you were correct that my only concern was the application of unloaded HT at switch on.

Hello,
the issue is not that simple to just use a series resistor with a silicon diode.
If we look at the right most graph here
https://www.jj-electronic.com/en/ez81-6ca4
we can estimate the dynamic resistance of one section of the tube to be approx. 25V/200mA = 125 Ohm.
It appears that a series resistor of this value would do as we need. But it won't. Why?
If we use a 2x250VRMS transformer with the HT winding impedance of 150 Ohm (as dedicated for EZ81 for this voltages) the resulting inrush current with a Si diode with a 125 Ohm series resistor can approach approx. 1.41*250V/(150+125)Ohm =~ 1.28A.
This is too much for typical reservoir capacitor, it will be a short life for it in such a circuit.
The inrush current with a tube rectifier on the other hand, will start with zero amperes rising smoothly to the nominal load current as the filament acquires the nominal working temperature; the reservoir capacitor will be fine - no surge current at all.
So replacing the tube rectifier with a solid state one calls for some additional circuitry mimicking the "slow start" behaviour of tubes.
One solution would be a series NTC thermistor in series, inserted where appropriate. Another one would be an RC delay in connection with a high voltage power transistor (like those used decades ago within horizontal deflection output stages of CRT TV sets).

[Gabe001]

Nemo, I think you need to also include the ESR of the transformer in your calculation.

If the primary transformer resistance is, say, 50ohms, and each HT winding is 150 ohms (assuming centre tap), this also serves to limit the inrush current by contributing an additional 50+150/2=125 ohms (I think). So total resistance 250ohms

The thermistor is a good idea, particularly when the transformer ESR contribution is low, but I don't think it's absolutely necessary in this case. Happy to be corrected if wrong.

[nemo_07]

Originally Posted by Gabe001

Nemo, I think you need to also include the ESR of the transformer in your calculation.

If the primary transformer resistance is, say, 50ohms, and each HT winding is 150 ohms (assuming centre tap), this also serves to limit the inrush current by contributing an additional 50+150/2=125 ohms (I think). So total resistance 250ohms

The Rt value as given in the tube data https://www.jj-electronic.com/images...opis_001_v.jpg is meant to be the impedance of supplying HT secondary (as seen by the single diode of the tube), so it still incorporates the impedance of the primary of the transformer (as well as the internal impedance of AC power lines).

Originally Posted by Gabe001

The thermistor is a good idea, particularly when the transformer ESR contribution is low, but I don't think it's absolutely necessary in this case. Happy to be corrected if wrong.

As short calculation presented above, the inrush current will be as stated. The only question is how much surge current will the reservoir capacitor tolerate for given span of life.
Manufacturers give often max. ripple current values for 1000h of life. Taking n identical paralleled capacitors divides the ripple current in a single part by n. Higher cost, weight and size, but this could work as well.

[murphyv310]

Hi.
The addition of a timed relay actually makes the situation worse in some situations as not only do you have the inrush current to charge the reservoir capacitor but also the heated valves that also increases the load when the relay closes. Yes a surge limiter is still required but why go to all this faffing around to save on one paltry EZ81, It's not as if it's a rare valve.

[ms660]

Originally Posted by nemo_07

Hello,
the issue is not that simple to just use a series resistor with a silicon diode.
If we look at the right most graph here
https://www.jj-electronic.com/en/ez81-6ca4
we can estimate the dynamic resistance of one section of the tube to be approx. 25V/200mA = 125 Ohm.
It appears that a series resistor of this value would do as we need. But it won't. Why?
If we use a 2x250VRMS transformer with the HT winding impedance of 150 Ohm (as dedicated for EZ81 for this voltages) the resulting inrush current with a Si diode with a 125 Ohm series resistor can approach approx. 1.41*250V/(150+125)Ohm =~ 1.28A.
This is too much for typical reservoir capacitor, it will be a short life for it in such a circuit.
The inrush current with a tube rectifier on the other hand, will start with zero amperes rising smoothly to the nominal load current as the filament acquires the nominal working temperature; the reservoir capacitor will be fine - no surge current at all.

Only if it's not hot switched, the valve data linked to below for the EZ81 shows that the surge current should be limited to 1.8 Amps, the valve data also suggests that the resistance of the diode at that current is approx. 47 Ohms (60*1.414/1.8) but I suspect it might be higher than that insofar as they might have include a safety margin factor in the surge limiter graph, that said the diode resistance at 1.8 Amps will be considerably lower than it's resistance at say 200mA:

https://frank.pocnet.net/sheets/030/e/EZ81.pdf

Lawrence.

[E93AFAN]

Gabe001, love the 3:3, made a mono version about 50 years ago when EL84s and EF86s were relatively cheap and transformers equally so, mine was paired with the 2 valve preamp, which for a bodger like me was easy to make. It was really good fun, it is a pretty good bit of kit too so my compliments to you for a very well executed project, I'll bet it sounds as good as it looks.

I didn't realise I was opening such a complex subject, I would be a careless handler of the truth if I pretended to understand the complexities of the subject as carefully covered by MS660, Gabe001 and Nemo_07, but nonetheless I thank you all for taking the time to reply.

The zener diode + resistor option sounds simplish, but as I have always stressed in my posts I am an experienced bodger with very poor maths (the result of a lacklustre 50's education) and so I'd struggle to go about calculating a suitable resistor + zener pairing with any confidence of success.

My lasting impression is that it might be far simpler just to find space under my chassis for a heater transformer and so despite my earlier comment about sucking it and seeing I shall now go down this route.

Thank you once again, I am grateful for your support in making my decision and potentially making an expensive mistake. I wish you all a very happy day.

[Gabe001]

Personally, I think you'll be absolutely fine with a solid state rectifier + resistor. The resistor will be around 125-150 ohms. It will get hot, so use 10w or greater.

In addition to the this resistor, the transformer resistance (impedance) will also contribute to the total resistance and decrease the current surge. Use a good quality capacitor from a reputable supplier, preferably one rated for a long life with a high ripple current.

If you want to be extra safe, put a thermistor between your on-off switch and the transformer primary. A cl140 should work well for you as long as your total capacitance is less than 150microfarad.

[Wellertcp]

I've done this several times; using silicon, plus a series resistor, and I've never had any problems. Dare I say - don't overthink it too much, just go for it.

[stevehertz]

Originally Posted by Wellertcp

I've done this several times; using silicon, plus a series resistor, and I've never had any problems. Dare I say - don't overthink it too much, just go for it.

I think you've hit the nail on the head there Weller.

[kalee20]

Originally Posted by nemo_07

If we use a 2x250VRMS transformer with the HT winding impedance of 150 Ohm (as dedicated for EZ81 for this voltages) the resulting inrush current with a Si diode with a 125 Ohm series resistor can approach approx. 1.41*250V/(150+125)Ohm =~ 1.28A.
This is too much for typical reservoir capacitor, it will be a short life for it in such a circuit.

In practice, it should be fine!

The limiting current for a capacitor is a thermal one. Sure, 1.28A ripple current continuously will not be good. But 1.28A charging current for a fraction of a second, how much heat is going to be generated? How much electrolyte is going to be vapourised and force itself out of the sealing bung?

If you think in those terms, you'll see there is no problem. The ripple current will be specified for steady-state conditions, and a few cycles of charging current is no way going to approach steady-state.

[nemo_07]

Originally Posted by kalee20

In practice, it should be fine!

The limiting current for a capacitor is a thermal one. Sure, 1.28A ripple current continuously will not be good. But 1.28A charging current for a fraction of a second, how much heat is going to be generated? How much electrolyte is going to be vapourised and force itself out of the sealing bung?

If you think in those terms, you'll see there is no problem. ...

Maybe you are right and my view is too conservative.
The reason for my concerns was a report I spotted once, where a short series of relatively high surge currents (on the mA/uF basis) resulted in irreversible increase of leakage currents in electrolytic caps (high value, low voltage types of renowned vendor were tested). As this worsened with every next surge, the phenomenon can be interpreted as sort of minor cumulative structural damages (sort of small local spots with process impurities or whatever).
No conclusions in this respect were given. Nevertheless, it is reasonable to assume that such degradation impacts adversely the reliability/life expectation.

While vendors do not supply a specification like maximum inrush surge current rating for given count of switch-on cycles, it remains a guess work. The idea was, that if we limit the switch-on surge current to the ripple current rating or less we should stay on the safe side.