power supply 1-wire remote sense

That looks just like a three-wire resistance thermometer connection.

John

Yep, and the downside is, any contact resistance in connectors or mismatch in current-delivery wire resistance does it in. Point-of-load regulation is cheaper (wire has real cost) and has better latency and bandwidth than digitizing-at-the-source. Is this a motor control, or heater, or some other milliseconds-don't-matter system?

Similar concept, an assumption of equal wire resistances.

It's a programmable DC power supply, and the customer has requested remote sense.

Here's the opposite extreme of remote sense:

That has some nice features.

I remember the early 80s when Atari used to use sense lines to monitor the power supply output at the game board.

Many smoked card edge connectors later they gave up on the sense lines. What was happening was the ground connections to the game boards deteriorated at the same rate as the +DC ones and this confused the sense(s) so they would allow more current to flow - and toasted the card edge connectors and the plugs.

We had a lot of problems with ground/common connections on games designed in the 70s and 80s.

I assume your design allows for this - I was just doing some reminiscing...

John :-#)#

Yes, rs can be connected wrong and make a power supply go nuts. The dual ADC allows us to compare the output and the rs feedback and see if they make sense.

I'd prefer fewer ADCs (thinking of these as feedback elements into an op amp-style regulator) because of their delays. Ideally, just program a reference, and apply gain to that, with suitable filter caps on the output. MHz power currents come from the capacitors, kHz power currents come from the amplifier, and a single ADC can alternate reading the reference and the output; using three ADCs seems overcomplex. If you just integrate DC errors (P-I-D style) using the digital stuff, the delay isn't an issue.

As for 'make sense' decisions: overcurrent or overvoltage would usually not wait for a smart digital chip's decision, but do something appropriate with a protective relay, circuit breaker, fuse... Undervoltage though, that's where a sense wire solution shines (take a millisecond, but get the value RIGHT before a guy with a meter sees the problem).

In message-id <t6nt3e$7bp$ snipped-for-privacy@dont-email.me

posted Thu, 26 May 2022

12:50:54 -0000 (UTC) John Dope stated:

Yet, since Wed, 5 Jan 2022 04:10:38 -0000 (UTC) John Dope's post ratio to USENET (**) has been 61.2% of its posts contributing "nothing except insults" to USENET.

** Since Wed, 5 Jan 2022 04:10:38 -0000 (UTC) John Dope has posted at least 1898 articles to USENET. Of which 173 have been pure insults and 989 have been John Dope "troll format" postings.

The Troll Doe stated the following in message-id <sdhn7c$pkp$ snipped-for-privacy@dont-email.me:

And the Troll Doe stated the following in message-id <sg3kr7$qt5$ snipped-for-privacy@dont-email.me:

And yet, the clueless Troll Doe has continued to post incorrectly formatted USENET articles that are devoid of content (latest example on Sat, 18 Jun 2022 12:53:36 GMT in message-id <kVjrK.298018$3%% snipped-for-privacy@usenetxs.com).

NOBODY likes the John Doe troll's contentless spam.

This posting is a public service announcement for any google groups readers who happen by to point out that John Dope does not even follow the rules it uses to troll other posters.

QIMDPZ0HUYw9

The power supply dynamics (voltage regulation, current limiting, output impedance, slew rates) will be done in an FPGA. The FPGA reads ADCs and generates PWM into the output fets.

The ADCs are ADUM7703, an isolated delta-sigma thing. It makes essentially a 1-wire random noisy duty cycle whose average value tells the input voltage. It needs to be decimated, digitally lowpass filtered, in the FPGA, and we have choices. We can get 160 KHz bandwidth at 15 bits effective a/d resolution, or 20 bits at about 50 KHz, both overkill for a power supply. They are $6 each, which isn't a big deal here.

Three ADCs isn't complex, it's actually simple.

If the customer doesn't elect to enable remote sense, they get a free high-res DVM. Either way they can verify the actual voltage at the load. They do high-value tests (like blowing blades off jet engines to test containment) and want to be sure everything works.

I've noticed that some engineers will go to great lengths, and add a lot of complexity, to save trivial parts cost. Like building a fancy charge-pump power supply just because they can, instead of buying a $3 dc/dc converter, to cite one fictitious example.

You idiots have mostly ruined usenet with your endless hen-clucking.

Your are either bots, or have astounding boredom tolerance. Of course you can't design electronics; your personalities are all wrong.

John Dope stated the following in message-id <svsh05$lbh$ snipped-for-privacy@dont-email.me

posted Fri, 4 Mar 2022

08:01:09 -0000 (UTC):

Yet, since Wed, 5 Jan 2022 04:10:38 -0000 (UTC) John Dope's post ratio to USENET (**) has been 61.2% of its posts contributing "nothing except insults" to USENET.

** Since Wed, 5 Jan 2022 04:10:38 -0000 (UTC) John Dope has posted at least 1906 articles to USENET. Of which 173 have been pure insults and 994 have been John Dope "troll format" postings.

The Troll Doe stated the following in message-id <sdhn7c$pkp$ snipped-for-privacy@dont-email.me:

And the Troll Doe stated the following in message-id <sg3kr7$qt5$ snipped-for-privacy@dont-email.me:

And yet, the clueless Troll Doe has itself posted yet another incorrectly formatted USENET posting on Sat, 18 Jun 2022 13:53:54 -0000 (UTC) in message-id <t8kldh$cug$ snipped-for-privacy@dont-email.me.

This posting is a public service announcement for any google groups readers who happen by to point out that John Dope does not even follow the rules it uses to troll other posters.

1uwkxZECbDJh

+1, and I'd add the need for constant attention seeking. Severe mental disorder. J

What? A relay is too fast to wait for a digital chip to control it??? A RELAY?

The input is a switchmode supply, intrinsically power limited by its nature, and only responding to logic input after finishing a current pulse. It's not lightning protection here, just craziness of the switching logic to be considered.

I'd not trust the switching logic more than a relay. Latency in sampling, and decision logic, and switch period, may not be negligible.

A much bigger concern is that none of the usual distributors seem to have ADUM7703s in stock. It looks like a nice device - I could use a few if only I could buy them! John

I guess I'll use diffamps into a grounded ADC instead of truly isolated ADCs. The ADUM is great for hanging across a current shunt, but there are not-quite-isolated ways to digitize currents too.

Looks like we can still buy opamps and resistors! Not much else.

<snip>

"Neon, argon, krypton, and xenon are obtained from air in an air separation unit using the methods of liquefaction of gases and fractional distillation."

Everybody liquifies air and sells liquid nitrogen. Elaborating the the plant a bit to pull out the noble gases isn't rocket science. Russia and the Ukraine may have tooled up for cheap mass production, but replacing them isn't going to be a big deal. It would probably just mean taking old kit out of mothball storage.

That also avoids problems with multiple grounds. Connecting a BNC cable to something at the load end could trash the 1-wire sense measurement.

We're exploring new micros, because the new products are all we can get. We're also having to make DFN-to-SOIC turret-style adapter boards because some asshole bought up the world's supply of ADA4899s in the sane packages, and is selling them for $15ish instead of $5ish.

Using the adapter saves spinning the board, and allows us to go back to normal whenever that blessed state arrives. (Hopefully it isn't like the starship waiting for the lemon-scented paper napkins in The Restaurant at the End of the Universe.)

Cheers

Phil Hobbs

M**** scrooed us badly, so we had to make 3000 of this: