Michell GyroOrbe Pabst AC Motor Power Supply

[User211]

For whatever reason using 2 AD9833s just doesn't work despite trying loads of pin configurations and much interaction with Grok.

It's also a wiring mess. So given that one can be made to work and using shorter wires than in the shot above, I'll try and do what the original PS did and just use a cap for the phase shift.

Simpler and way less messy plus performance should still be good.

[User211]

Shaping up just need to connect amp output terminals and mount everything to the chassis. That 50V 22uF Rubycon - it is probably stupid to use it given its age, so I will order a new one from Amazon. It goes across the amp's output terminals.

TBH I don't rate Grok much with electronics. It keeps coming out with utter rubbish. If you don't have at least a basic understanding of what you are doing you simply cannot rely on it to design something that will work. You have to see through it and steer it in the right direction. You get gold from it sometimes, though, so it has been a worthwhile activity. I don't think I would have done this without chatting to it.

[Wakefield Turntables]

WOW! What an amazing thread, many many thanks for sharing Justin.

[Barry]

What is 'Grok'? You refer to it as "it", so is it an AI design app?

[User211]

What is 'Grok'? You refer to it as "it", so is it an AI design app?

It's Elon Musk's AI built into X formerly known as Twitter.

Like Chat GPT if you have heard of that.

It's generally quite impressive on all sorts of subjects. Ironically it is remarkably good at disassembling X posts and exposing them for distorting facts.

Precisely the antidote for EU claims of spreading misinformation.

I therefore think due to its inclusion the EU has zero case for suing Elon for billions of dollars.

[User211]

WOW! What an amazing thread, many many thanks for sharing Justin.

Don't thank me too soon this could be a complete lemon!

I think I am into doing a bit of DIY I have quite enjoyed this thus far. It's just finding the time away from work to do it that is a challenge.

If anything about it doesn't work fixing it shouldn't be much of an issue cost wise.

My wife was out a couple of days ago and came back with an interesting book.

Here's a couple of pics.

That looks like a cool amp. Look at the date stamp on the book.

It ain't gonna drive Apogees, but I bet that sounds interesting.

Most powerful DIY SET amp of the time. Pic is a bit blurred, so I'll give it away - 1934.

[User211]

I reckon that 2 AD9833s really ought to work. Also, the three SPI leads should be removable once the AD9833s are programmed. They are really only used for clocking in the data that is the program.

That means the wiring loom with these shorter leads should be fine.

I will return to it once this simpler config is proven. Probably just bad pin soldering on my part. Maybe.

[User211]

We really do want that second AD9833 working.

Why? Because common 22uF 50V caps are typically +/- 20% tolerance. That's pretty poor, really.

Why does it matter? Because you will only get maximum torque from the Pabst when the phase shift is bang on 90 degrees. 85-95 degree phase shift will be acceptable according to Grok.

So does a 20% out of spec cap translate into a 20% phase shift? No, thank god. Grok reckons it could be out by about 10%, though, which isn't what we want.

A way to do OK is to buy a few caps, attach the PSU to the Pabst, and use a scope to find which cap gives the most accurate phase shift. Pain in the arse. Or maybe use a breadboard.

Note the Pabst will only run at its smoothest i.e. with the least vibration when the phase shift is bang on (in theory - there may be manufacturing tolerances in the motor itself I guess).

With the 2nd AD9833 working we can get the phase shift truly accurate in circuit with the Pabst by merely reprogramming the AD9833 to get the best in circuit phase shift. Zero soldering in this case. And it should never drift like a cap might over the years.

Apparently 5% tolerance caps are available but I had a quick look and couldn't find any at 22uF 50V.

Waiting for the Nichicons I ordered but they are 20%.

So turntable PSU do have some depth to them when it comes to getting a good engineering result.

[User211]

One more point.

I was talking to a Michell rep at the Bristol hi-fi show this year about my busted PSU.

He said mine should output 22V. Been wondering why but I noticed the Pabst is designed to run at 19V to 24V, so I guess Michell went for somewhere in the middle of that range.

[User211]

Feel free to correct me I am just blagging my way through this trying to work out what's best as I go on.