Let me know what you think of Jan's circuit, I like that as it doesn't need to tap off the 6.3vac windings, makes for an easier time with PCB traces, also no relays, it uses a MOSFET for a switch.
Fixed resistors would be nice, but in order to keep with the original spec of being more versatile, eg: 2.5-12.6VDC @ 4.5A, we'll need to keep the pot in. I think a switchable array of resistors would be a bit much for a footprint. That said I am open to expanding upon this feature, unless I'm missing something there's a decent variety of tube filament voltages that could be powered by this hence the need for 2.5-12.6 and not strictly 6.3/12.6?
I've picked out a tentative MOSFET and redrawn the circuit according to ra7's helpful comments. Tentatively looking at STP11N65M2 as it's around the same price as that BUT11A, and seems to have good heat dissipation and low rdson. I'll be posting a v1.1 schematic with that soon. I've also swapped the original loop breaker circuit into a CL-60 and 22nf 300VAC+ Y2, changed large cap footprints to 22mm diameter, and changed the solder points to 2 or 3 terminal 5.08mm pitch terminal blocks.
For that 5W resistor, the original BOM says 100-220R or remove if using a choke, the schematic shows it as 1K. I have always interpreted this as the R in a CRC filter and could be sized anywhere from 0.1-whatever as long as users calculate proper resistor wattage for the amount of voltage drop desired across it. Overall I'm still working on cleaning up the BOM to PR03's, changed parts, etc.
I have had a look at Jan's HV delay circuit - both the original one in 2014 and the updated one in 2019. The big problem is the need for a pre programmed PIC controller IC. This comes as part of the DiyAudio store kit with the pcb and fets. A nice design though. Maybe one can program the chip themselves when I read the article from 2019 - do you know for sure on this?
I am OK with the pot to adjust the output voltage.
I would probably use a choke instead of R4, all good with the other mods.
I am OK with the pot to adjust the output voltage.
I would probably use a choke instead of R4, all good with the other mods.
You are right, I only glanced at the 2019 doc from Jan and didn't catch that you have to program the chip. Maybe the ESP design is ideal as it doesn't require anything over the top like that, I'm hoping to avoid relays for longevity.
Using a good quality relay will not be a problem, especially when using 2 contacts in series as per the ESP article.
The ultimate circuit is the one that works with the minimum amount of components.NE555 is unbeatable for delaying anything.What's the reason to have an IC that has auto reset? That's just one more problem if PIC supply gets unstable .lm317 will always allow for the perfect fillament slow start with just 2 more components(a 10...22uf cap and a resistor)...but a cap multiplier is just good enough.
The word " ultimate" is used in excess to justify all sorts of expensive complications. If you're comming from a fix and repair background you know that the simplest solution is the best one too.
The word " ultimate" is used in excess to justify all sorts of expensive complications. If you're comming from a fix and repair background you know that the simplest solution is the best one too.
This project is sort of a ‘clone’ with modifications. The name “Ultimate…psu” comes from the original designer and should stay that way.
Keeping the HV delay circuit simple with minimal parts count is something I agree with.
The filament psu section works fine, no need to make changes there with added parts.
Keeping the HV delay circuit simple with minimal parts count is something I agree with.
The filament psu section works fine, no need to make changes there with added parts.
We aren't going with any programmable chips at this point, right now looking at a variation of Pete Millett's muting delay circuit. eg: a 555 timer to control a 250v relay with the contacts in series like Rod Elliott's circuit shows. I've drawn up Rod's but since it uses a 16dip CMOS timer, I feel it will have a larger footprint, and I think I can have a smaller footprint with a riff off Pete's circuit. Essentially the top half of the attached circuit as I don't think we need the footprint of the LED portion unless other people want me to include that. Where I am unsure is, can I tap off the rectified filament DC current that has ground elevated to 1/4 B+ voltage, and run the high voltage delay circuit with an elevated ground? That would save on components in the circuit.
What the hell do you need 2 x ne555 for?
If you chose dc fillament and /or slow start filtered , you shouldn't care about anything as long as you couple only the B+through relays and have different secondary windings for filament and HV supply.The fillament lifting is better done by connecting the fillament through 100k...10Megohm to 1/4 V B+
You can lift that fillament even if it's AC and the effect still works.In fact dc fillament doesn't completely solve the noise because that is in the picoampers range where no regulator really is noiseless, fillament lifting is much more potent.
If you chose dc fillament and /or slow start filtered , you shouldn't care about anything as long as you couple only the B+through relays and have different secondary windings for filament and HV supply.The fillament lifting is better done by connecting the fillament through 100k...10Megohm to 1/4 V B+
You can lift that fillament even if it's AC and the effect still works.In fact dc fillament doesn't completely solve the noise because that is in the picoampers range where no regulator really is noiseless, fillament lifting is much more potent.
As I said in my post, just the top half, which if you notice, only uses a single 555 timer. Per the schematic I attached as an example where I'm just using the top half as a starting point, you'll see that the bottom half with the second 555 is only if you wish to have an LED, which I do not. I believe I mentioned in my previous post that I was not interested in the LED portion of the example I posted. You're welcome to read up on that particular circuit on Pete Millet's website.
I think you should be alright with the delay circuit with elevated gnd at 1/4 B+, as long as you don't connect LV gnd to HV gnd anywhere. Otherwise you will need a separate supply.
Also I note the use of 2 relays with SPDT contacts, you only need 1 relay with DPDT contacts as far as I can see. Depends what the relay contacts are switching.
Also I note the use of 2 relays with SPDT contacts, you only need 1 relay with DPDT contacts as far as I can see. Depends what the relay contacts are switching.
Good to know on the ground thanks, one relay is the plan, Pete's schematic is originally tailored for signal output hence 2 relays, as you said, we only need one. 🙂
Good work, Bloqhed. Here is an interesting circuit - it counts mains cycles for a power on delay - no microprocessor
http://www.next.gr/circuits/Simple-B-delay-circuit-for-tube-power-amp-l38806.html
http://www.next.gr/circuits/Simple-B-delay-circuit-for-tube-power-amp-l38806.html
Thanks I will take a peek at this one.
Been a bit busy at work this week, but I've still made some good progress. I need to refine the high voltage regulation a little bit and I should have a new schematic to post then, hopefully in the next couple of days. In parallel to the schematic I'm also working on fine tuning things like resistor footprints to maximize available board space.
I would simply float a UA723, add a zener follower to power the chip, and You will have a cleaner B+ and better regulation for about the same price. expect at least 2uV noise level.
I really like the sound of that excessively low noise level. But here the novice in me screams. Do you have any example schematics I can look at to get some ideas and understanding? I'm a learn by example type.
This is'nt a floating design but here is one Ive worked up in both Spice and breadboarded, it works great under no load and 10mA load. it may need some refinement, but not much I would think.
Attachments
