• WARNING: Tube/Valve amplifiers use potentially LETHAL HIGH VOLTAGES.
    Building, troubleshooting and testing of these amplifiers should only be
    performed by someone who is thoroughly familiar with
    the safety precautions around high voltages.

Current measurement in choke input supply


2005-08-21 10:46 pm
Hi there,

question: In a choke input supply, would it make sense to measure the DC current in the 1st choke and assume that this is the average of the total current in the transformer winding?

As far as I understood the theory, the choke current vs. time should consist of a DC component and a (symmetrically) superimposed AC part...

Background: I almost killed my tube amp by operating it during elevated temperatures last summer. We had 40°C and above outside, and in our flat directly below the roof temperatures reached >30°C. While the transformer was chosen with very narrow margin in terms of power requirements and operating temperature, everything went well until this summer. The 10°C increase in ambient temperature pushed the transformer over the edge, a plastic interlayer insulation started melting and turning brown while releasing an unpleasant odor.

As there is not enough room to throw in a larger EI core, I am going to switch to a toroid. But before ordering the expensive made to order piece, I'd really like to confirm the theoretical current consumption to have enough margin this time.



2006-04-17 6:27 pm
For a choke input filter the RMS current in each half of the transformer secondary may be taken approximately as 0.75 times the direct current to the load.

As far as I understood the theory, the choke current vs. time should consist of a DC component and a (symmetrically) superimposed AC part...

That depends on the design of your transformer. Only with half wave rectification there will be a DC component on the secondary when the amount of DC ampère turns is high. Temperature rise is dependend upon the cooling area, the total loss and the ratio of iron to copper loss. I've once made aluminium plates and bolted them at both sides of a mains transformer. It ran a five degrees cooler but it's better to replace the transformer with a more potent unit.
Last edited:
I'd suggest using PSUD2 to set up your situation - it will indicate the rms current in each full-wave half-winding (assuming that is what you want). It will also indicate the choke AC and DC levels, if that is what you want. You can also vary items like loading to see what influence that has.

I'm surprised you had an insulation failure - do you know the specs of the insulation used in the transformer? Perhaps it was a cheaper item with a low insulation temperature rating - the ratings can vary by 30 to 70C between common levels, with up to a 220C level. If the insulation can take the temp, then it doesn't really matter what temp the transformer gets to (apart from nearby parts).


2005-08-21 10:46 pm
Secondary AC RMS current will almost always be significantly higher than average DC current, even for a choke input supply. 'disco' suggests a ratio of 1.5.

So the secondary AC current will be 1.5 times the average DC load current, but it flows only half of the time in each half of the secondary in a FW rectifier design. So each half of the secondary should at least be able to deliver 0.75 times the load current.

By designing the secondaries with the expected load current in mind I should be on the safe side then?

A simulation in PSUD requires detailed knowledge of the total load current, which I am not sure of. There's no single load, I have the driver stage, output stage, various bleeder resistors and a relay. There might be a significant error if I just sum up the theoretical currents.

That is why I was looking for a convenient point to measure the total current drawn from the secondary. Of course the best place would be the secondary, but the current waveform there is beyond what my simple DMM is able to measure...

Assuming the choke feeds a first filter capacitor, then you could insert a temporary current sense resistor in the negative lead from that first filter capacitor that goes to all your equipment (but not the power transformer CT) - that resistor would be at ground potential, and be mainly DC with low ripple.

Similarly, you could insert a temporary current sense resistor in the CT link to the negative terminal of the first filter capacitor - that resistor would be at ground potential, and be mainly DC (depending on your choke size) and your DMM may be able to separately identify DC and AC voltage levels across that sense resistor.


2016-06-07 6:50 am
Start with an amplifier that has a 3 wire grounded power cord powered by a 3 wire grounded mains outlet (I hope your amplifier has this).

Then, check that the amplifier, as designed, already has the the following three items connected together:
Chassis, B+ first Capacitor return, and the ground of the 3 wire power cord.
Now you can insert a 1 Ohm resistor between the first Capacitor negative return lead
and the junction of the already connected Chassis and 3 wire power cord ground.

Then if your scope has a 3 wire grounded power cord going to the same 3 wire grounded mains outlet, then you should be able to see the voltage waveform across the resistor.

If the scope is digital, it may have measurement features that will enable the automatic measurement of the rms value of the voltage across the resistor.
Because the resistor is 1 Ohm, the amperage will be the same number as whatever the scope voltage rms measurement reports (Volts # = Amps #).

The waveform will be able to show you the peaky nature, or the smooth nature of your first capacitor current versus time.
Your mileage may vary versus another persons amplifier; but you will know what Your amplifier capacitor current is like (like if he uses a different choke than you do, even if his schematic is the same).

Things that do affect the Capacitor current value, shape, etc.
1. Cap input filter
2. Choke input filter, with a choke inductance equal to or greater than the 'critical inductance'.
3. Choke input filter, with a choke inductance less than the 'critical inductance'.
4. A Choke that can Not take the current swings causing it to saturate.
5. A Choke that Can take the current swings and Not saturate.
6. Vacuum Tube Rectifiers.
7 Solid State Rectifiers.
8. Choke DCR
9. Power Transformer Primary DCR, and Secondary DCR.

10. And you asked about the current in the secondary windings:

A bridge rectifier will have all the current in all the secondary windings, all the time.

A full wave rectifier with center tapped windings, will have all the current in the half winding for only half of the time (the windings alternate the current duties).

As you can see, there are many reasons some people seem to get very widely differerent results.
Often, it is due to the items 1 - 10 above.
Last edited:


2006-04-17 6:27 pm
There might be a significant error if I just sum up the theoretical currents. That is why I was looking for a convenient point to measure the total current drawn from the secondary. Of course the best place would be the secondary, but the current waveform there is beyond what my simple DMM is able to measure...

Ordering a new transformer for this amplifier, I'd start with the (theoretical) circuit total consumption and multiply this value with a factor 1.2 to 1.5 to help regulation and cool operation. When this theoretical information is not available you just put a low value resistor (whatever value but not higher than 10R) in the HT and apply Ohms law ;) When built to order, the (assumed experienced) manufacturer keeps AC ripple current in mind so you won't have to worry about that. A good design runs luke warm with nominal power dissipation.
Last edited:
Choke input supplies are excellent. (many of my amps have & have had them), and were used commonly on amps and modulators that have large variations in current.
Don't try to reinvent the wheel, it has all been done perfectly well before.

A good CIF can have a suprisingly small choke, and small filter C.
It's really astonishing how well they filter out noise and hum.

Buy a NOS one from UTC, and see how much voltage drop you get across it under maximum output.
You need a "Swinging choke", nothing else.

Swinging chokes are rarely available today, as everyone seems to want CLC, which can be suprisingly inefficient.
CIF mutter & groan, give excellent line stability, and generally speaking a lot stress on the rectifiers.
You can even run them in a negative return line, rather than run them "LIVE" at the hot end.
So the secondary AC current will be 1.5 times the average DC load current, but it flows only half of the time in each half of the secondary in a FW rectifier design. So each half of the secondary should at least be able to deliver 0.75 times the load current.
You got the right answer but the wrong method. A true choke-input power supply* has a power factor of approximately unity. That means the RMS current in the transformer secondary is equal to the DC load current.
But, in a two-diode rectifier the RMS current in each half of the transformer winding is not simply half the total; it is equal to the total RMS current divided by 1.4 (sqrt[2]). RMS is funny like that.

In other words, if your amplifier draws 100mA DC, the total transformer secondary current will be 100mA RMS for a true choke-input supply, and 71mA RMS in each half of the secondary if you're using a two-diode rectifier. So you got pretty much the right answer by the wrong method...

*But some people get confused about this; do you really have a choke input rectifier, or just a regular smoothing choke?
Last edited:
Choke input has ratio of DC to AC RMS current of unity only when the current drawn is significantly greater than the minimum current needed to ensure 'choke-input' action. At lower currents the ratio increases a little. Modelling is one answer; a true RMS meter is another method. The various places to insert a sense resistor mentioned by various people above will not get you very far without an RMS meter - most DMMS will not do this.
At lower currents the ratio increases a little.
True, but I'd argue that is irrelevant since the currents are, as you say, lower. The ratio isn't likely to exceed 2 (PF = 0.5) so as long as the choke doesn't drop out of regulation before the DC current drops to half its nominal value, you're still safe. And who would design a choke input filter that did that? Most valve amps don't even have that range of current variation.
Last edited:
All real waveforms are continuous. Yes, the current waveform will not be particularly peaky but it will not be a sine wave so an ordinary DMM will not give the right results. However, if the PSU is running well above the minimum current than the AC contribution will be small so an error in measuring it will have less effect.
..who would design a choke input filter that did that? Most valve amps don't even have that range of current variation.

Having a little experience with CIF, I can honestly say with the kind of amp that demands them (typically AB2 class), I can assure you the current variations are enormous from say 70m/a for a pair at idle to as much as 300m/a+ under full load.

On one of mine it dragged down the HT on it from 520V to a little over 400 on full load.
When you add up power transformer winding losses, rectifier voltage drop under full load and the choke filter internal resistance, it's easy to understand why lots of older valve amps suffer from compression... ie. very non linear.
The only thing to say, with a C-L-C the effect is actually considerably worse,with hum pumping thrown in, but lots of people can't hear it.

It's why most professional modulators and high powered transmitters/rigs opted for choke regulation.
If you choose the right swinging choke inductance, the right voltage and current, you get really impressive regulation with low noise.

I fixed my line sag in an unorthodox way by preceding the valve rectifier by Si rectifiers.
The valve rect anodes were then connected together and the whole thing was paralleled.

It's suprising to see how much LESS line sag you get when the 2 paralleled rectifier diodes are on all the time....doing just DC, rather than having to cope with a sine wave - on/off 25hz signal per diode.

If you regulate screen grids so they are rock solid, it's amazing to see how much low distortion power you get out (as much as DOUBLE!).
On lots of the good 40s and 50s amps, they did C-L-C on the screen grid + AF amp supply, and CIF on the main 500-700V HT lines, with 2 seperate rectifiers, one for each secondary winding.
Last edited:
So the valve recto was just acting as an expensive resistor?

EDIT: It provides a soft start, I suppose.

More than a soft start.
Anyone who knows a little about the subject knows valve rectifiers do a great job of removing line noise - mostly the higher frequency mush which can also come from diodes made of sand.

If you then stuff what comes out of there through a swinging choke, you get an excellent PSU, in fact in that particularly case, the hum was so indetectable, you had to stick your ear by the speaker...and wait for it, that was a smoothing cap of only 8-16uF total.

I couldn't hear any hum at all, but there was a tiny little hiss.
Like I say, why try to reinvent the wheel?
The guys from the 30s and 40s knew their stuff.
I wanted to add.
I am a strong believer in removing hum & noise with voltage regulators.
Why try to do with C-L-C filters what you can do so much better with regulator valves or simple gas shunt stabilisers?

A more important reason can be the much much higher sensitivity of screen grids to PSU induced noise and hum.
Many screen grids have so much gain, you can drive them instead of the control grid, so it's pretty obvious that ripple and variations to the anode voltage (which are cancelled out in a P-P amp), have very little influence on the noise performance of the amp.

On top of that, to get anything like a steady regulator at anything over 600V you need to have an input voltage to the regulator of at least 750-800V, & a well insulated heater supply for the regulator - which rapidly makes things more complex.
It's another very good reason for using 2 different independent power supplies for the 2 rails, because they do rather different things and are different impedances.

On one amp, I used a well known regulator triode to derive the AF amp supply direct from the main 670V line rather than opting for the original 335V screen supply, because the screen supply dragged insane amounts of current out of that one.
No amount of chokes and decoupling could have stopped that g2 line sagging therebye introducing non linearity/HF dropout under high loads, into the preamp and driver stages. (another big weakness in the old DTN Williamson/classic British EL34 based designs)...
Last edited:
Dunno … I have read thru the thread, and I'm left with the feeling (as often is the case), that different people have had different results; that there are advantages and disadvantages to CLC and LCRC designs, respectively, and that there is a whole lotta backpedalling to the halcyon years of Olde, an so on.

Perhaps 'cuz I don't have a “dog in the fight”, I see this: choke-input power rectification and at least one CLC stage following of further filtering does a good job, within the confines of its nominally high open-load to full-load power droop merit. The droop is real, and having either a high enough valued choke, and a particular amount of minimum current draw more or less gets such a topology to its almost-invariant plateau voltage. Having a large enough reservoir capacitor following further isolates B+ voltage fluctuation under AF load; moreover, one can make the argument that such a choke-input filtering section is almost ideal for solid-state rectification, since a choke blocks all the HF noise that our nearly-magical pieces of purified sand produce.

Except for the voltage drop, itself.

The simple-as-can-be “solution” of course is to put a modest-valued capacitor up front, before that choke input. Capacitor captures the rectification peaks, drooping every inter-peak cycle in proportion to the current draw and the condenser's capacitance. The output B+ voltage goes up substantially since the average of the first-capacitor's voltage is much higher than the raw, just-rectified transformer output.

Now, according to 6V-heater's evidence, such a power supply has a droop far larger than a choke input supply. I've personally not encountered this, actually. Going from a 70 mA load to a 300 mA load over a short interval definitely causes any inductor in the CLC path to have to “ramp up” its current flow. Moreover, when the increased load is suddenly returned to near-quiescent (or worse, near-zero), well … the big old L in the CLC has to put on the brakes. The C₂ of the C₁LC₂ combination in turn “takes it” and rises in voltage beyond quiescent.

Clearly none of these strategies are without quixotic fault.

At least to me though, the 'simplest of the simple' is to steer away from huge pieces of iron and large capacitors, and take the operating maximal range of the load into consideration, then design backwards. Backwards with an X between the 2 C's. CXC for X=regulator. Regulators, unlike the steampunk flywheel inductors, react nearly instantly to drooping output, increasing flow from the “left-hand-side” C₁ as needed, and in solid proportion to the increased load. They do exactly the opposite, just as fast, with a suddenly lowered current demand.

Moreover, they're almost comically simple devices to hook up these days, especially in the era of 1200 V MOSFET transistors. IF our design goal is to (1) cut ripple a LOT, (2) stabilize B+ below 1% droop under high load, (3) get job done inexpensively, and (4) reliably, too…, then the series MOSFET regulator is simplicity itself. No Zeners, gas discharge voltage references or anything like that really needed, since (5) Line voltage independent output was not among the criteria.

A $100 plus item (the choke) is replaced by less than ten bucks of parts, and with a 5 kilo savings in mass. They do need heat sinks and turn-off-protection back-emf rectifiers tho, to be properly designed.

The only real reason I'm putting this out there, is because while we've had a most scintillating discussion of (not quite) calculating proper choke-input choke and load values, and an equally lengthy discussion of the relative merits of The Old Timer's high-tension regulation-and-filtering ideas, in the end I think we might all agree upon this:
A Perfect Real World Power Supply
• efficient
• reliable, safe, uncomplicated
• removes almost all rectification ripple
• doesn't droop between 0 and 300% of nominal load​
Right? And that particular combination can be had, rather surprisingly easily, with CLC(XC)n designs. The ()n business is reflecting on 6V-heater's observation that separately regulated power supplies are vital for output-finals screen grids, for input-section anode supplies and the rest. Whether you use silicon rectification, vacuum rectification, or the cute-little-cheat of “both” is almost immaterial. And the L in that CLC business is smallish. There more for filtering all of the line cord's rectified harmonics, than acting as a big old flywheel. 1 Hy is quite enough for any vacuum-tube amplifier supply in this CLC(XC)n configuration.

C1 = 22 µF, L1 = 1 Hy, C2 = 220 µF, (X = 1200 V MOSFET, C3n = 47 µF)n

I've implemented just thus, and had excellent results. Being somewhat more irrational about having precisely regulated voltage absolute values, I use a stack-of-Zeners for conjuring up the reference voltages. That way, the amplifiers are equally tweak-free on 110 V power supplies, as they are 125 V mains. Or 220 V vs 250 V in Europe.

Finally I apologize if the above opinion was unwarranted in the hallowed halls of the Tube forum.

Just Saying,
GoatGuy ✓