I'm working on a 2x swappable octal tube cathode biased Class A push-pull relatively low B+ (goal of 270 VDC) and power output guitar amplifier design, with selectable pentode, UL40, and triode operating modes, and I'm currently modelling the power supply in PSUD2. Here's what I have so far:
Hammond 370KX power transformer, 250-0-250 @ 322 mA, 5.0 V @ 6.0 A, 6.3 V @ 6.0 A
5AR4/GZ34 valve rectifier
Hammond 193H filter choke, 5.0 H, 200 mA, 65 Ohms
Kemet hi-temp, long-life, low ESR Al electrolytic capacitors, 450V
Hammond 1650PA output transformer, 6.6 KOhms primary, UL40 taps, 16/8/4 Ohm secondary
I am currently modelling this using a assumption of 2x6L6GC tubes drawing a maximum current of 165 mA @ 270 VDC, as per the datasheets.
My power supply filters look like this, with a pi filter input and successive RC filter stages:
C1 47 uF, 1.2 Ohms
L1 5.0 H, 65 Ohms
C2 22 uF, 2.4 Ohms
I1 165 mA (the two 6L6GC tubes), V=279.14
R1 1 KOhm
C3 10 uF, 5.7 Ohms
I2 2.4 mA (12AX7 phase inverter,) V=271.88
R2 2.2 KOhms
C4 4.7 uF, 11 Ohms
I3 2.4 mA (second 12AX7 preamp stages), V=261.27
R3 4.7 KOhms
C5 2.2 uF, 24 Ohms
I4 2.4 mA (first 12AX7 preamp stages), V=249.99
So far, everything looks great, except one thing. the choke current. the choke I've selected is rated at 200 mA, and once the amp settles about 200 mS after power on, it's drawing about 172 mA. but during startup, PSUD2 shows the choke current peaking at about 625 mA.
Now, I am not exactly an electrical engineer, so I don't know if I really need to worry about this, but even if I change to the larger 193L choke, which is rated at 300 mA (which I may do anyway, given what I am seeing about EL34 plate current in Class A on the datasheets, but given that I am planning on a 1650PA OT, which is 6.6 KOhms primary impedance, I'm not certain if I need to worry?), I still get a current peak at startup that exceeds the rating.
Should I be worried? What else can I do?
Hammond 370KX power transformer, 250-0-250 @ 322 mA, 5.0 V @ 6.0 A, 6.3 V @ 6.0 A
5AR4/GZ34 valve rectifier
Hammond 193H filter choke, 5.0 H, 200 mA, 65 Ohms
Kemet hi-temp, long-life, low ESR Al electrolytic capacitors, 450V
Hammond 1650PA output transformer, 6.6 KOhms primary, UL40 taps, 16/8/4 Ohm secondary
I am currently modelling this using a assumption of 2x6L6GC tubes drawing a maximum current of 165 mA @ 270 VDC, as per the datasheets.
My power supply filters look like this, with a pi filter input and successive RC filter stages:
C1 47 uF, 1.2 Ohms
L1 5.0 H, 65 Ohms
C2 22 uF, 2.4 Ohms
I1 165 mA (the two 6L6GC tubes), V=279.14
R1 1 KOhm
C3 10 uF, 5.7 Ohms
I2 2.4 mA (12AX7 phase inverter,) V=271.88
R2 2.2 KOhms
C4 4.7 uF, 11 Ohms
I3 2.4 mA (second 12AX7 preamp stages), V=261.27
R3 4.7 KOhms
C5 2.2 uF, 24 Ohms
I4 2.4 mA (first 12AX7 preamp stages), V=249.99
So far, everything looks great, except one thing. the choke current. the choke I've selected is rated at 200 mA, and once the amp settles about 200 mS after power on, it's drawing about 172 mA. but during startup, PSUD2 shows the choke current peaking at about 625 mA.
Now, I am not exactly an electrical engineer, so I don't know if I really need to worry about this, but even if I change to the larger 193L choke, which is rated at 300 mA (which I may do anyway, given what I am seeing about EL34 plate current in Class A on the datasheets, but given that I am planning on a 1650PA OT, which is 6.6 KOhms primary impedance, I'm not certain if I need to worry?), I still get a current peak at startup that exceeds the rating.
Should I be worried? What else can I do?
Last edited by a moderator:
Relocated to Instruments & Amps. There will be someone here with the needed insight.
Oh, thank you! I did not realise I was in the wrong forum!
A turn on current surge in to a choke in that application is not a concern at all. PSUD2 can't simulate the saturation response of the choke anyway, so what you see is not what you'll get.
One risk could be the hot turn-on peak current through the valve diodes. PSUD2 indicates that peak is nearly 3A. The datasheet allows up to 3.7A. if you sim a lower inductance then the initial peak doesn't increase much, so that is not too concerning.
Another risk is exceeding the continuous peak diode current, as you are getting close. If you wanted to push the loading current up to say 200mA then you would be sitting pretty much on the peak limit.
The OT datasheet doesn't indicate primary winding dc resistance, so some doubt about what plate idle voltage would be.
One risk could be the hot turn-on peak current through the valve diodes. PSUD2 indicates that peak is nearly 3A. The datasheet allows up to 3.7A. if you sim a lower inductance then the initial peak doesn't increase much, so that is not too concerning.
Another risk is exceeding the continuous peak diode current, as you are getting close. If you wanted to push the loading current up to say 200mA then you would be sitting pretty much on the peak limit.
The OT datasheet doesn't indicate primary winding dc resistance, so some doubt about what plate idle voltage would be.
Last edited:
Maybe try the sim with C1 and C2 values swapped?
The first cap causes most of the inrush current.
The first cap causes most of the inrush current.
From what I can see, the only way I'd get to 200 mA load is if I swapped out all the preamp tubes for 12AT7s or 12AU7s (~10 mA per triode), or if the EL34 datasheets' numbers for plate current in Class A SE pentode mode are directly convertible to Class A PP pentode mode; from what I have seen of other tubes whose datasheets do specify Class A PP numbers, that isn't necessarily the case. KT88s are a complete crapshoot, unless I can find some hard data on operating characteristics in that voltage range.
I'm trying to avoid having to go to 2x5AR4s. I'm already bummed out enough that I won't be able to freely swap in a 5Y3GT, because there's just not enough current there.
6L6GC numbers are given in Class A PP at 250 and 270 V into a 5KOhm primary. EL34s generally run about 3.4 KOhm, as I understand, but Mercury Magnetics states the early EL34 Marshalls actually used the 6.6 KOhm primary OTs from the earlier KT66 amps, and they sound great, so I'm going to go with that.
Just making a comment if you happened to try a KT88 or 5881 or ...
Are you needing to use valve diodes? A hybrid with valve diode in the B+ line may allow 5Y3GT with plates paralleled.
Were early EL34 Marshalls operating class A ?
Are you needing to use valve diodes? A hybrid with valve diode in the B+ line may allow 5Y3GT with plates paralleled.
Were early EL34 Marshalls operating class A ?
Valve rectifiers have a certain behaviour that is considered desirable for instrument amplification in some contexts. If I design for a valve rectifier, I can capture that sound, I can swap in different valves for different characteristics, and I can also use a solid-state valve replacement in the valve socket. If I design for a solid-state rectifier, I don't have the same options.
EL34 Marshalls were Class AB fixed-biased.
> choke current peaking at about 625 mA
Bah. It would take an hour for the choke to start burning. This peak is way-way-way shorter.
(Near 100X overcurrent I would wonder for windings flying off the core. There are some pictures of utility generator transformers exPLODed by a short. But you are not near that.)
> I did not realise I was in the wrong forum!
I say it does not matter that it is a guitar amplifier; this is Power Supply.
> 12AU7s (~10 mA per triode)
AU7 "can" suck 10+mA. But if you use them with the audio-usual 100K-50K resistors and 300V supply, you can't get 6mA uselessly slammed and not over 4mA in a happy working condition. Anything over a Watt or so output completely dominates all the "small tubes" put together. (Unless it is an Organ with rows of oscillator/dividers.)
> 6.6 KOhms primary
And 280V supply?
The maximum plate current, ideal power devices, slammed to Square Wave output, is 169mA. No real tube can suck this hard, say 150mA. Add 10% for G2 and 10mA for small bottles, is 170mA.
Sine Test current will be lower, around 0.63X. (0.707*0.9) 108mA.
Attached is my derivation of effective resistance load on the power supply, for *ideal* devices, several topologies, several waveforms. Knowing R you can get I if you know V. However a power amp IS a resistor-like load, not a constant current drain. If you double V, I will essentially double.
No real tube will swing 100% as assumed here, 90% or 80%. So R would be higher. OTOH we have G2 current and small-stuff, so R will be lower. These values can be used directly in PSUD's simpler "R load" with better accuracy than is warranted for the poor specs we have on transformers.
Bah. It would take an hour for the choke to start burning. This peak is way-way-way shorter.
(Near 100X overcurrent I would wonder for windings flying off the core. There are some pictures of utility generator transformers exPLODed by a short. But you are not near that.)
> I did not realise I was in the wrong forum!
I say it does not matter that it is a guitar amplifier; this is Power Supply.
> 12AU7s (~10 mA per triode)
AU7 "can" suck 10+mA. But if you use them with the audio-usual 100K-50K resistors and 300V supply, you can't get 6mA uselessly slammed and not over 4mA in a happy working condition. Anything over a Watt or so output completely dominates all the "small tubes" put together. (Unless it is an Organ with rows of oscillator/dividers.)
> 6.6 KOhms primary
And 280V supply?
The maximum plate current, ideal power devices, slammed to Square Wave output, is 169mA. No real tube can suck this hard, say 150mA. Add 10% for G2 and 10mA for small bottles, is 170mA.
Sine Test current will be lower, around 0.63X. (0.707*0.9) 108mA.
Attached is my derivation of effective resistance load on the power supply, for *ideal* devices, several topologies, several waveforms. Knowing R you can get I if you know V. However a power amp IS a resistor-like load, not a constant current drain. If you double V, I will essentially double.
No real tube will swing 100% as assumed here, 90% or 80%. So R would be higher. OTOH we have G2 current and small-stuff, so R will be lower. These values can be used directly in PSUD's simpler "R load" with better accuracy than is warranted for the poor specs we have on transformers.
Attachments
Last edited:
- Status
- Not open for further replies.