Need help with 600v+ PSU

[ecc83]

Well, if the bleed resistor sees 600+ then that 450V cap is in trouble!

No it won't.

Fusing big trannies is always a problem. With totally discharged caps and stone cold heaters the initial current will be huge, a 5A T might survive it 20times but maybe not 100?

In a production scenario you would take a transformer and see what sort of fuse blows with various secondaries shorted at say-& +10% mains in ( and for all world voltages!). I presume this T cost you a bit and you don't want to risk that!

If you find spurious fuse blowing a problem you might consider a few Ohms in each bridge leg to limit turn on current? They would also save the caps in the event of a rect' punching thru'.

Dave.

[defaced]

If so, it's going to be expected to dissipate 6.137W!

Even if it sees half of that, it still dissipates 1.5w. I'd increase the resistor value. Since it's just there to provide a path to ground, I'd go as large as possible while still keeping a reasonable discharge time (if that matters to you). 220k is what I am using in a build I'm working on and even with a lot of capacitance for a guitar amp, it's dead in less than five minutes. If you use 220k, dissipation drops to about 1/2 watt.

[andrew_k]

Dave: I'm having a little trouble understanding, but they say the only stupid question is one you don't ask, right?
This amp will never have both the filaments and HT fire up cold at the same time (because no one but me will ever use it and theoretically I'll remember to use standby..), so it'll only be the cold heaters or empty caps loading the PT. Under that scenario is 5A ok? 5A is the highest commonly available slo-blo in the 3AG form factor here, hence my desire to use it.

defaced: Good point, I will increase the values. Thank you.

[ecc83]

Hi Andrew,
Well, if indeed the amp will never leave your greasies you don't actually need a fuse at all, you can do WTF you like! But don't die! In such a chunky amp a 13A plug fuse would pop in the event of an HT short I am pretty sure.

BTW if you are talking 20mm fuses they can be got up to 20A for Mwave ovens, think I have some 7's somewhere.

Yup Mikes point is well taken, just a balance between discharge time and juice "wasted", my mental "map" just remembers that I have some 68k 3W metal films somewhere!

Dave.

[andrew_k]

BTW if you are talking 20mm fuses they can be got up to 20A for Mwave ovens, think I have some 7's somewhere.

So would 10A be a better choice? There's a local supplier with those as well.
I would prefer to get the fuse selection sorted soon as the rear faceplate is about to be remade (the first one was back to front -- power socket on right instead of left etc. I truly am a moron sometimes ) and I have "5A SLO-BLO FUSE" written above the fuse holder. I'll change this to 10A I think.

At least the front came up nice

[ecc83]

Ha! Ha!

I wouldn't worry that much about the legending! Dymo to the rescue!

In 50yrs in electronics I can assure you "they" have all got it "wrong" sometimes and us service techs used to get "kits" of upgraded fuses and sometimes mains filter caps to mod various products as they found spurious fuse popping in various countries.

Middle of the road devices,radios, audio systems (lowish power) TV's all these were fine but for very low drain devices, vcr's e.g were often fused at a perfectly sensible 250mA but these blew for no apparent reason and were uppped to 315mA.

Big audio amps were also something of a trial due mainly to the emergence of torroidal trannies! And fuses were often only fitted in mains inputs to comply with a particular countries (often restrictive then) "safety" laws.

Bit of a Black Art fusing.

Dave. Nice BTW.

[andrew_k]

Bias supply

The attached image shows my attempt at adding a bias supply.

It combines elements of the Ampeg V4 bias supply and the dual bias pots of the AX84 October.

The only problem is I don't know how many volts I need on the bias supply. With 646V on the plates, I'd want to read 37.9mV at the test points for 6550's and 45.5mV for KT88's. I attempted to find the formula for the required bias voltage in RDH4 but after about 45 minutes my brain shut down

[ecc83]

EEKK!
That's never going to go more than about 50v negative of 600!

I always think this bias "pinch" circuit is a cop out but if you cannot find it in your heart to fit another tiny transformer at least try to design a pretty failsafe supply.

Use 2 caps off the bottom winding each of which can supply all the bias current and go for a decent voltage, say 70V and feed that into an LM317 or even just an MJE340 and a stack of zeners in its base, then pot off from that.
Remember this is class AB, that secondary is going to wobble a bit!

'Twas not long ago I had to replace a mains tranny in a 50+50watter due to loss of just that simple cap' fed bias supply. The 6550's were pretty forked as well!

Dave.

[andrew_k]

HA!!! Of course it will!!
This is what I get for doing this stuff before my morning coffee.

I have a 24V@2A transformer spare. How about I run that into a voltage doubler?

[andrew_k]

Also Dave, would you be so kind as to please tell me the formula for calculating how many volts of negative bias supply I need to achieve a set bias point? (eg: 39mV using the "1Ω to ground" method)
Thank you!

[ecc83]

"I have a 24V@2A transformer spare. How about I run that into a voltage doubler? "

ha, HA! NOW you are talking Andy! Don't go for any fancy capacitor coupled gizmo. 2 1/2 wave diodes on one leg give about 60V and at a decently low source resistance.

It is always good to keep the grid resistance of power valves to deck as low as possible.

Dave.

[ecc83]

Sorry Andrew, no can do, otherwise the punters would not need amp techs like me to bias their kit!

All you can do is take the valve data figure as a start point and set the bias "cold" to max neg then back it off for the wanted Ik.

I know I keep giving you more work but have you considered a dc balance pot* and an HF balance trim from the phase splitter? I assume you want to extract maximum acoustic damage from this beastie!

*Tip, if you do this, set for minimum AUDIBLE hum and not perfect dc balance.

Dave. (err.. Forget the HF trim if you don't have a 'scope.)

[andrew_k]

ha, HA! NOW you are talking Andy! Don't go for any fancy capacitor coupled gizmo. 2 1/2 wave diodes on one leg give about 60V and at a decently low source resistance.

Excellent, thanks! I'll try and draw something up after work.

Sorry Andrew, no can do, otherwise the punters would not need amp techs like me to bias their kit!

haha, fair enough. I'll aim for -30V to -20V per tube as a starting point and work from there. Once it's fired up I'll be taking it to a great local tech for his expert eye and so I can measure the clean output on his scope+dummy load rig. Fingers crossed for crossing the 100W mark

[ecc83]

Sorry again mate!

Duncan amps data (googe) gives: AB U/L Va 553 Vg1 -75V! for 100W.

Even for 70W and Va 453 you need -59V that's for Ia = 2x50mA.

You might need a voltage tripler after all. I note the KT88 is almost exactly the same.

Dave.

[defaced]

My 2x KT-88 setup at 475v B+ needs about -60v on the grid to idle at ~58ma.

For your setup I think I'd be shooting for a bias supply that can do -85v or more so I have wiggle room for different grades.