PSU Design

There aren't very many switcher or linear reg chips that accept more than about 30 volts. Many, a lot less.

Can you post the old schematic?

I have that issue now, namely a +48 buss and a need for a bunch of small switchers. I like the old LM2576HV-ADJ - quiet and reliable - but it's big and switches at 52 KHz so needs a giant inductor and caps.

The 150-kHz and 250-kHz Simple Switchers are pretty good, and let you use much smaller passives. I use the LM2596 and LM2672. The genuine TI ones aren't what you'd call cheap.

I have a couple of TDS 784As, one on the bench and one in a rack next to it. The slow control response takes a bit of getting used to, but they're generally very good. 1 GHz, 4 GSa/s is pretty useful.

Of course their FFTs and so on are pretty slow, but they're probably running a 68020 or something inside.

Cheers

Phil Hobbs

The *old* schematic? What old schematic, John?

Oops, missed that link.

It is a real hairball. Do you really want to fix it?

I guess it's a 60 Hz transformer, not a switcher.

You'd have to know the load currents to redesign the regs.

Do the supplies really need remote sense? That shoots down a lot of simple approaches.

Color digital scopes are great. Fast, cheap, light, run cool, have storage and signal averaging, show the volts and time ranges, and none of ours have failed so far. And they show traces in color! Monochrome scopes confuse me now.

Ideally, yes. However, harsh experience has taught me that the next "phut!" is seldom far away. These components are now half a century or more old and even if they don't fail, their values drift over time, eventually out of spec so it's strictly a labour of love.

Yes, I mentioned in the description it's a linear PSU, but If I'd have posted the complete schematic that would have been far more evident. My bad.

Indeed, yes. However, I can get those from a fellow enthusiast who keeps data on that kind of thing.

I'm not sure, John. You're right, though, it would be much simpler if they were not needed.

I know, I know. I really should get one and it's not as if I can't afford it. But I'm kind of wedded to the CRT type for nostaligic reasons and the bond is very strong! You still rate Rigol highly? Any others?

but pass transistors can

I'm almost tempted to ask why the psu lines are interdependant.

That was one of the two questions I posed about this design. The only thing I can think of is they were trying to save a couple of windings from the transformer secondary to keep cost down.

We mostly buy Rigol lately. A few Teks; I really like the fully isolated TPS2024.

The Rigols have been great. My guys have developed a Python library to talk to them for automated measurements.

My bench scope is a 4-channel 500 MHz Rigol. We bought a few of the

I think the Rigols have got more expensive since some tariff law change.

I do love my ancient, maybe 80 pound, 11802 sampler.

They all seem to use one another. Startup must have taken a lot of thinking.

You'll get good help on the TekScopes folk on groups.io (the group split in two - try both):

I assume you already know about the TekWiki at

Clifford Heath

Hard to say. It's clearly a very old design, and the schematic you've posted isn't complete - and doesn't have a date on it.

There's no voltage reference shown anywhere on the part of the schematic you have posted, so trying to second-guess why the designer came up with what's shown here would be a waste of time.

The 130V output voltage isn't widely available. Mouser offers

When I need a semi high voltage output or input from HV and not too much dissipation, I just take an N channel FET in source-follower configuration with a zener diode (V = Vout + VgsTh) from gate to GND and a high-ish value gate to HV bias resistor. If independence of output voltage is important, you could just add more of these circuits.

boB

That kind of advice assumes a certain amount of electronic competence in the person being advised. You are talking to Cursitor Doom, who doesn't have it.

Linear Technology parts come with pretty fool-proof data sheets.

I don't see a lot of interdependence, just a lack of integration, which is typical of the date of mfr.

Do whatever you want, wiseguy. You'll find out.

RL

The CR things are crystal rectifiers.

In those days, D meant dynamotor, a type of dc/dc converter.

afford it. But I'm kind of wedded to the CRT type for nostaligic reasons and the bond is very strong! You still rate Rigol highly? Any others?

I really like my Keysight 3000 A 1GHZ scope but it cost a small fortune...Last year I got a Siglent sds1104x-e and am amazed at all the capability it has, for $500USD. Get the 100MHz version and apply the free upgrade hack that takes it to 200MHz. Scope has 30+ math functions, 4 channel, I2c/RS232/SPI/uart decoding etc. It doesn't have external sync.... There are Rigols in a similar price range with similar functionality (I think the Rigol has ext sync).

No external trigger connector? No big loss; they are usually terrible anyhow.

I usually trigger off channel 4. Much less jitter.

It's full of CR things. Not a D in sight.

Resistors are 4.7k, not the dreadful 4k7.

C985 is .0047; nanofarads hadn't been invented yet.

They do avoid 4-way connections, which people did in the days of hand-drawn schamatics, when connection dots would fall off the page.

What's going on with Q829?

I hate the upside-down ground rakes.

Not quite.

CR = CURRENT RECTIFIER

This design intent was to control startup sequence to ensure the highest voltage source enables the 50V which enables the 15V.

If the HV was slow in rising, some circuit that uses both supply voltages would see that as a polarity reversal.

For 3 terminal LDO's they mitigate this prompting users to add a reverse diode from input to output so if the higher input voltage goes down fast, the diode will protect BE junctions from excessive reverse voltage |Vr| > 5V.

It was a good question that did not deserve snarky responses.

Tony Stewart EE since 1975