Where did all my volts go ?

[IslandPink]

Ok, this is hurting my head but should be easy for clever people to solve.
Another power supply question :

I'm experimenting a lot with filament supplies at the moment , on driver valves, and in this case , on an output 300B valve.
So, I'm looking for a 5V supply and about 1.6A .
I have this passive supply built in a style promoted by Thorsten Loesch years ago on his 'Ladyday' amp modifications thread.

The circuit for that is below .
My version of it is subtly different but not much .
I have a screen shot from LTSpice attempting to model it .

So, if I feed it with 7V AC ( measured , meter ) into the diodes bridge, I measure only 4.65V DC out .
Spice suggests about 5.4V out ( with 9.8V pk-pk input )
PSUD, with only a choke in the upper rail, suggests over 6V out .

So were does it all go in practice ( chokes are about 0.4 ohm DCR ) .

[IslandPink]

Ooops, spice model :

ps. the diodes are actually 1N5822 schottky's in mine.
I'm expecting about 0.6V per diode .
About 0.4V across the resistor
Maybe 0.8V across the chokes ( let's say they are 0.5R ) .

[IslandPink]

The thing is, y'see, I'd quite like to buy a mains transformer for running GK-71's and GM70, but I don't have enough confidence in my calculations at present to put in the order ....

[IslandPink]

I notice the CMC is losing a bit - must be ~0.1R on each side, as I lose ( measured, real unit ) 0.3Volts from that . So I have 4.8v before the CMC .

I've used 0.065 ohms ESR for the caps, which is about average for a 10,000u/16V .

So overall , Spice Sim ( 5.3V with everything ) vs. reality ( 4.49V ) shows a missing 0.8V .

Both caps in the real supply are now new or new-ish .

[pre65]

Only 0.8V difference ? In my book that's accurate.

How many VA is the transformer ? I tend to overrate mine so they sag less.

[JamesD]

I would suspect that the impedance of the transformer itself is causing it to not be able to full recharge the caps on each half cycle i.e. the top of the sine wave is being chopped off by the current demand of the caps so that they are not reaching their full voltage..

J

[IslandPink]

The transformer is a fairly hefty Parmeko one where I'm using 2x 2A 6.3V windings for each supply module, so I thought I was safe on that .
I guess the voltmeter could be fooling me in what it's reading as '7V AC' depending on how the waveform looks ?

I think I'm close enough to spec my GM70 fil. transformer now - should have some margin with a 2x18V ( which will be over-volts typically ) for 20V out . Just debating whether to go 250VA or 300VA ....might be good to have the extra margin, for an extra £6 !

[pre65]

I used 18v 120va toroids (one per valve) on my last GM70 project.

They kept relatively cool.

[IslandPink]

Thanks Phil.
Maybe I'll be OK with a 250VA with 2x18V then . It does save me some space as the 300VA ones are 125mm dia !

[Nick]

The transformer is a fairly hefty Parmeko one where I'm using 2x 2A 6.3V windings for each supply module, so I thought I was safe on that .

Yep, but that will be rated for AC load, the DC supply will be pulling a big spike of current for a small percentage of the cycle, so what would be fine internal resistance for 2A AC may start to matter when rectifying to DC.

[JamesD]

resistance if futile!

Lower the impedance for impulsive tasks!!!

[IslandPink]

Ah yes, having looked at the current out of the transfomer in Spice, I see how short the duty cycle is in this case. I can see how that might affect the predictions . I suppose for starters I need to measure the resistance of the secondarie(s) and feed it into the Spice model , but I guess the real limit could be due to the core ?

Thinking ahead, also , based on the dicussions ( Nick, James ) we've had this week . If I was going to use a simple volts-reg supply for this and apply the virtual divider set-up, what's the device ( hefty ) that would be recommended for GM70-style currents ?

[Mike H]

Yes increase the resistance until it is more realistic. Don't forget primary has DCR as well. I assume you are using actual coupled coils in the sim to make the transformer?

[IslandPink]

I haven't got that far with the sim, Mike !
I didn't envisage doing that much work , just establishing some loss mechanisms , to enable me to judge what bits to buy for a GM70 supply .
I'll see if I can find time this week in between doing some hardware swaps, to improve the sim . Will it be easy to take the mains TX model we used on the T-Rex supply ? - I kind-of expect I'd need to measure primary and secondary inductance, plus primary and secondary DCR , for the Parmeko transformer, to get a useful model
Do you think the biggest contribution, in this case, is likely to be the secondary resistance ? - if so, I can measure, check that effect, and stop at that point .

[Mike H]

Good lord just plonk 2 coils on, L1 & L2, plonk on a K directive 'K1 L1 L2 1', as it's mains, make L1 say 100H minimum.

Secondary is then the Voltage ratio squared times the primary inductance.

Measure the DCR's and put them in the inductor properties.

The tricky bit is the regulation percent. Ideally want to know what this is but can guess at 5 - 10%.

The secondary DCR can be made to be this 'X' percent of the secondary Voltage, divided by the maximum current the thing should produce (Ohms law), the off load Voltage will be higher than spec., on load the Voltage will be dropped accordingly depending on the current draw. For example Danbury makes them to produce correct Voltage at maximum load (max. current).

If you're hyper fussy you could do things like tweak the primary inductance to get the mains side leakage magnetisation current loss, for example, but we're not studying the behaviour of the transformer in this case, so I wouldn't bother.

Anyway if you get it right you should end up with something which overall gives more realistic effects.

HTH