I see Two Circuits/What circuit is best? (linear bench PS and know how)

[001]

Hi!

This units looks similar. But not the same

One from 80`th and other is from 2000`th

That the REAL difference is?

Why it use different feedback architecture and ameter trick?

Is are still good?

[001]

The first

[001]

The second

[001]

no opinions? 

[T3sl4co1l]

Neither one is very good.

Tim

[bd139]

First one is waaaay too complicated. Second one smells of mastech. And that smell is of failed relays and destroyed 2n3055s and smoked target boards.

Look at the HP62xx supplies and E3630 if you want something that isn’t ass.

[001]

First one is waaaay too complicated. Second one smells of mastech. And that smell is of failed relays and destroyed 2n3055s and smoked target boards.

Sorry for my english

What is "too complicated" means? What about mastech (does it good or bad? why)

Look at the HP62xx supplies and E3630 if you want something that isn’t ass.

Thanx! Can U say why?

[bd139]

Better than my English and I’m English

Too complicated means that it has been over engineered. A power supply which is reliable and effective doesn’t need that many parts

Mastech power supplies have a couple of failure conditions. They are mostly reliable until they are not. What happens is the relay that switches the transformer taps dies after a bit and then the supply has to sink a lot more power through it than if the relay was working. This causes the pass transistors to blow up and go C-E short dumping the entire voltage on the pre regulated stage of the supply into your circuit. This happens a lot. I have fixed perhaps four units that have done that.

Why HP supplies? The design originated in the 1960s and been incrementally improved since then. That’s 50+ years of fixes in there . E3630 because it is three power supplies in the same box so you can see what they did to handle different voltage scenarios.

[Whales]

To go a bit further:

Too complicated means that it has been over engineered.

Too complicated = too many parts = too many things to break = unreliable

A simpler design is often a better design.



To understand why the two circuits you post are so complicated you need to understand the challenges the designers were faced with.  Examples:

• Old ICs such as the ua741 and LM301 have lots of limitations/problems
• Mass manufactured circuits are designed to be cheap, not well performing or reliable
• Each power supply has a different output range and capability

Sometimes a simpler/better design with less parts will actually cost more money to make, because the better design uses more expensive parts.  Sometimes.

[001]

Look at the HP62xx supplies and E3630 if you want something that isn’t ass.

Thanx

I see what hp6236b is a basic design without any preregulator -- more heat fo transistors
E3630 use triac preregulator instead relays. Is it good? What about switching hum and stability?

[bd139]

There’s a tiny bit of noise on the switching of the preregulators but it’s a design compromise to keep it in the form factor. If it didn’t have this it’d have to be twice as large or twice as hot. The E3630 doesn’t really even get warm flat out. Relays fail which is a big problem really. Stability is 100% fine - HP put a lot of design thought into compensation, transient response and power up transients.

A trick you can do with a shitty power supply is get a couple of 500m rolls of wire (aka huge low Q inductors) and stick a big cap on the end and try and charge it up. A couple of the cheaper brands (I’ve seen this on an HY1803D) will oscillate slowly at the power supply terminals. I haven’t made one blow up yet doing this but it’s fun to watch.

Personally I use TTi PL series power supplies but the HP ones are an excellent and reliable design that is fit for analysis.

[001]

A trick you can do with a shitty power supply is get a couple of 500m rolls of wire (aka huge low Q inductors) and stick a big cap on the end and try and charge it up. A couple of the cheaper brands (I’ve seen this on an HY1803D) will oscillate slowly at the power supply terminals. I haven’t made one blow up yet doing this but it’s fun to watch.

wow 
but why HY1803D so popular? what the way to modify it?

[bd139]

I wouldn’t buy one or modify it. Better spend money on a second hand Hp/TTi/Agilent supply.

[001]

I wouldn’t buy one or modify it. Better spend money on a second hand Hp/TTi/Agilent supply.

It is not easy if You far from US/Europe

In fact all of linear supplies looks similar
The same opamps transformer and power transistor
So it looks not difficult to buid one  for fun

But what schematic is best for DIY clone? 

[bd139]

Fair point.

I’d clone an E3620A myself. Schematic here: https://www.eevblog.com/forum/testgear/agilent-e3620a-power-supply-no-schematic-available-says-agilent-support/

[001]

Fair point.

I’d clone an E3620A myself. Schematic here: https://www.eevblog.com/forum/testgear/agilent-e3620a-power-supply-no-schematic-available-says-agilent-support/

Thanx!
Why You prefer "A" instead "B"?
How looks Your clone (for mechanical and thermo reasons)?

[bd139]

There is no E3620B (yet)

I haven’t cloned it myself but that’s where I would start. It would take a lot of experimentation to get it built though and would likely cost more than buying one anyway.

If faced with the task I would probably buy a mastech and some big Zener diodes and a bunch of fuses to protect my circuit from the power supply 

[rstofer]

Is that multi-winding transformer used in the HP design actually available?  I can't read the parts list in the .pdf (oddball viewer, I guess) so I can't see a part number.

It would seem to me to be just as difficult to get that transformer into Antarctica as an entire power supply.  I suppose an alternative would be multiple transformers.

One question re: the first design: Could I realistically recreate the PCB?  For me, the answer is no.  I have neither the time nor the patience to take on such a project.

[001]

One question re: the first design: Could I realistically recreate the PCB?  For me, the answer is no.  I have neither the time nor the patience to take on such a project.

It is simplest and lazy part of work 

[not1xor1]

Fair point.

I’d clone an E3620A myself. Schematic here: https://www.eevblog.com/forum/testgear/agilent-e3620a-power-supply-no-schematic-available-says-agilent-support/

IMHO the triac preregulator is an overkill for such a limited amount of power.
A two winding transformer with a mosfet as a switch would have been more than enough.

A couple of months ago I tested (just in the LTSPICE virtual world, where burnt silicon doesn't smell  ) a similar design with MOSFET switches rather than triacs.

The main problem is that it is hard to find a 3 winding transformer.
I wonder how difficult it would be to unwind the secondary of a toroidal transformer and then rewind it in 3 separate parts .
Large transformers are probably more manageable as they have a larger hole and require less turns per volt.

[bd139]

The E3620 is designed for entirely passive cooling so it works out.

Agree with transformer. You can use more than one trandformer if you need to. Might be easier to do that off the shelf.

If you look at some of the newer units they are exactly the same design with mosfet pass transistor and DAC control + ADC sampling. Don’t fix what ain’t broke.

[001]

two winding transformer with a mosfet as a switch would have been more than enough

Can You post some schematics?

[rstofer]

The E3620 is designed for entirely passive cooling so it works out.

With sizable cooling fins!

I decided to look on eBay and the E3620A goes for around $250.  I can't build it for that!

I imagine shipping would be an issue.

[bd139]

They’re pretty expensive new. Around $800.  I couldn’t build it for that with the time involved.

[not1xor1]

The E3620 is designed for entirely passive cooling so it works out.

With sizable cooling fins!

I decided to look on eBay and the E3620A goes for around $250.  I can't build it for that!

I imagine shipping would be an issue.

on ebay.it (where .it means Italy) they are sold for around 100€ (HP, Agilent)

now I've spotted a HP3624... they have been sold for around 130€ lately...
it is quite old and huge and I think there are a lot of old capacitors to replace and probably the fan too

[not1xor1]

two winding transformer with a mosfet as a switch would have been more than enough

Can You post some schematics?

here is a proof of concept with 2 mosfet switches (but it would work with just 1 + a diode though with a bit less efficiency):



here are the resulting traces:



this might work both for halving the power transistor dissipation and to switch (via an external switch) between half voltage full current mode and full voltage .62 derated current mode.
That is if you have a 15+15V 150VA transformer you could use a switch to make your power supply work as 0-15V/0-5A PSU or as 0-30V/0-3A PSU.

[001]

WOW 

[macboy]

Fair point.

I’d clone an E3620A myself. Schematic here: https://www.eevblog.com/forum/testgear/agilent-e3620a-power-supply-no-schematic-available-says-agilent-support/

+1
The E3620A is fantastic. Super clean, low-noise power. Voltage is very stable from no load to full load, it has no ripple on the output and output noise must be measured in microvolts. Microvolts.  It is orders of magnitude cleaner power than anything else on my bench.  To top it off, the output is well-behaved upon turning on or off the mains power switch. Many other brands still don't know how to pull off that simple, necessary feat. It's a very good design, worthy of study and imitation. I really like the idea of putting the control sections' internal "ground" reference at the positive output terminal (not the negative one). The output transistor and the current sense shunt are up there on the positive output, so this makes so much sense.

The preregulator isn't complicated, it is just a comparator that turns on the TRIAC or doesn't, based on the output voltage, providing one of two voltages to the regulator.  It is not a phase-controlled variable output voltage style of preregulator like what is used on some linear high power designs.  The preregulator on the V2 side is similar but provides one of four voltages, resulting in lower power dissipation on that side.