From: "Harvey White" <madyn@...>
To: [email protected]
Sent: Thursday, April 2, 2020 9:54:54 AM
Subject: Re: [HP-Agilent-Keysight-equipment] HP 3455a DVM repair

I'd be glad to help if I could, but I don't have one.? What I've said is
based on the 3456As that I have.? Take a low frequency square wave, say
1 second on one second off.? Put that to the input of the meter.? The
manual should show you the A/D converter and the inguard processor (IIRC
it's inguard), You should be able to see that signal at the a/d inputs.
You should be able to identify the outputs of the inguard processor,
look to see if anything is changing there.? Then try the outguard
processor, that's the one that drives the display.? Again, you're
looking at board signals, not signals going so much between chips.

What I suspect, as do others, is that the inguard processor may not be
talking to the outguard processor.? The inguard processor may be
"broken" as a guess.? That the display changes with range is good.? That
it's all zeros may be that it's simply getting all zeros.

Look at the actual boards and check the signals there.? You don't need
to start looking at chips until you figure out what's going on.

Typically, the inguard processor is run from an isolated supply (and ALL
the voltages must be exact, had that problem with a -30 volt supply in a
3456), the input attenuators, AC converter, ohms current supply, A/D and
that processor are all run from that supply.? Then the output of that
processor goes through some sort of isolation network rated for many KV
to the main processor.? HP used a pair of highly insulated transformers,
just a few loops of wire and a core, to get a clock and data signal over
in the 3456A.? Likely they're doing the same thing here.

I had a bad inguard processor in one of mine.

Harvey


On 4/1/2020 8:19 PM, Bob Albert via groups.io wrote:

Harvey, my main problem is identifying the various parts.? The manual
isn't the easiest for me.? Perhaps you can walk me around to the
various places where I can make measurements.? I certainly have no
shortage of sources of dc or ac of any reasonable frequency and amplitude.

I did measure the 10V reference with my 3456A at 10.00007 I think.
(not sure how many zeros)? So what we have here are all the pieces of
a good meter if I can make it play.

Bob

On Wednesday, April 1, 2020, 05:09:17 PM PDT, Harvey White
<madyn@...> wrote:


If I might:

generally, a signal goes through attenuators, then is converted to
digital.? This is done in a floating processing environment called (by
HP) the inguard processor.? That is optically coupled to an "outguard"
processor, which is the main processor driving the display.? Any break
in the chain can result in zeros.

Firstly, I'd take the signal from the inputs, then see how far it
gets.? If you can see anything changing in the digitally coupled
outputs (optically isolated), might be working.? You'll need to check
on both ends of the optical isolators.? That's assuming they have
them.? The 3456 has two very isolated transformers that need to be
checked.? What I found is that one of the processors itself was dead
when I was trying to resurrect some of them.

So the bottom line is to try to follow the signal. An AC signal would
be easiest but you'd have to tell it that it's really DC, otherwise
you to the AC converter.? If you should try to do an AC signal into
the DC input, what you're looking for is the presence or absence of a
signal (to a point where the circuitry gets confused and/or ignores
it) and after that, the signal isn't there.? At that point, you need
to match the AC signal to an AC input, or DC to DC.

This is based on my experiences with a 3456, so YMMV

Harvey


On 4/1/2020 1:36 PM, Bob Albert via groups.io wrote:
Well Chuck, I don't know what to do next with this beast.? It acts as
though it works but the display gives nothing useful.? Yet the display
electronics seem to be working okay.

If I put a signal in, it ignores it. Open or short on Ohms, voltage or
no on volts, gives zeros and no autorange activity.? So perhaps it's
an input circuit issue.? In Test I get error 11.? I press test
repeatedly and it gives error numbers 10, 9, etc down to 0.

The manual sucks.? It tells me to look on module such and such but
nowhere does it seem to show me which module is where.

Bob

On Wednesday, April 1, 2020, 10:22:56 AM PDT, Chuck Harris
<cfharris@...> <mailto:cfharris@...> wrote:


Hi Bob,

I don't see how you can come to that conclusion.

The 3455A is highly reparable, and should be well
within your abilities.? Lots of people have fixed
their own successfully.

What you won't be able to do is send its calibration
module out for calibration.? You will have to do that
yourself.? But, it is not any more difficult to
calibrate than the 3456A.

It is not my meter of choice, as I greatly prefer
the 3456A, but for many, it is a useful addition to
their labs.

-Chuck Harris

Bob Albert via groups.io wrote:

? I don't see a whole lot of encouragement here regarding the

possibility of getting my 3455A working.? Maybe I will just put it
aside and wait for a scrap unit to swap out boards or parts.? My 3456A
still works so I am grateful for that.

Bob
? ? On Wednesday, April 1, 2020, 06:40:17 AM PDT, Chuck Harris

<cfharris@... <mailto:cfharris@...>> wrote:

? The 3455A was a dinosaur when it was first released.

The 3455A and 3456A were sold side by side, with the
3455A being about twice as expensive as the much more
capable 3456A.? The 3455A's interesting feature was
it was supposed to be cheap to calibrate. All you
had to do to do a complete calibration was to unplug
the module in the back panel, and replace it with a
recently calibrated module.? The idea was you could
order up a calibration by mail, and return your old
calibration module as a core... never having to send
the full meter in for calibration.

Interesting idea, but surely not worth twice the price
of a 3456A.

I say it was a dinosaur, because it used all small
scale logic rather than large highly capable hybrid
circuitry, like the 3456A.? It should be much easier
to repair, but it should also be much less reliable.

And, that is my experience with the two units.

-Chuck Harris

paul swedberg wrote:

With respect to the 3455 and 3456 the difference is night and day.

I have

numbers of 3456s and was given a seriously troubled 3455. Quite a
interesting meter.
Replaced at least 11 actually malfunctioning chips on the logic

board. The

logics simple so easy to verify the chips were bad. No shot gunning
involved. Never really did get it working but ended up with a different
3455 that needed help and the AC board and it is in fine shape. With
respect to the old 3455 it has both a logic and voltmeter issue.

The 3455

uses an unusual processor and teh DVM uses the same proc and a

masked eprom

that is known to go bad. It acts as a state machine.
The 3456s are losing there eproms and copies can be found on the

net. Thank

heavens.
Regards
Paul
WB8TSL

Harvey, my main problem is identifying the various parts.? The manual isn't the easiest for me.? Perhaps you can walk me around to the various places where I can make measurements.? I certainly have no shortage of sources of dc or ac of any reasonable frequency and amplitude.

I did measure the 10V reference with my 3456A at 10.00007 I think.? (not sure how many zeros)? So what we have here are all the pieces of a good meter if I can make it play.

Bob

On Wednesday, April 1, 2020, 05:09:17 PM PDT, Harvey White <madyn@...> wrote:

If I might:

generally, a signal goes through attenuators, then? is converted to digital.? This is done in a floating processing environment called (by HP) the inguard processor.? That is optically coupled to an "outguard" processor, which is the main processor driving the display.? Any break in the chain can result in zeros.?

Firstly, I'd take the signal from the inputs, then see how far it gets.? If you can see anything changing in the digitally coupled outputs (optically isolated), might be working.? You'll need to check on both ends of the optical isolators.? That's assuming they have them.? The 3456 has two very isolated transformers that need to be checked.? What I found is that one of the processors itself was dead when I was trying to resurrect some of them.

So the bottom line is to try to follow the signal.? An AC signal would be easiest but you'd have to tell it that it's really DC, otherwise you to the AC converter.? If you should try to do an AC signal into the DC input, what you're looking for is the presence or absence of a signal (to a point where the circuitry gets confused and/or ignores it) and after that, the signal isn't there.? At that point, you need to match the AC signal to an AC input, or DC to DC.

This is based on my experiences with a 3456, so YMMV

Harvey

On 4/1/2020 1:36 PM, Bob Albert via groups.io wrote:

Well Chuck, I don't know what to do next with this beast.? It acts as though it works but the display gives nothing useful.? Yet the display electronics seem to be working okay.

If I put a signal in, it ignores it.? Open or short on Ohms, voltage or no on volts, gives zeros and no autorange activity.? So perhaps it's an input circuit issue.? In Test I get error 11.? I press test repeatedly and it gives error numbers 10, 9, etc down to 0.

The manual sucks.? It tells me to look on module such and such but nowhere does it seem to show me which module is where.

Bob

On Wednesday, April 1, 2020, 10:22:56 AM PDT, Chuck Harris <cfharris@...> wrote:

Hi Bob,

I don't see how you can come to that conclusion.

The 3455A is highly reparable, and should be well
within your abilities.? Lots of people have fixed
their own successfully.

What you won't be able to do is send its calibration
module out for calibration.? You will have to do that
yourself.? But, it is not any more difficult to
calibrate than the 3456A.

It is not my meter of choice, as I greatly prefer
the 3456A, but for many, it is a useful addition to
their labs.

-Chuck Harris

Bob Albert via groups.io wrote:
>? I don't see a whole lot of encouragement here regarding the possibility of getting my 3455A working.? Maybe I will just put it aside and wait for a scrap unit to swap out boards or parts.? My 3456A still works so I am grateful for that.
> Bob
>? ? On Wednesday, April 1, 2020, 06:40:17 AM PDT, Chuck Harris <cfharris@...> wrote:
>
>? The 3455A was a dinosaur when it was first released.
>
> The 3455A and 3456A were sold side by side, with the
> 3455A being about twice as expensive as the much more
> capable 3456A.? The 3455A's interesting feature was
> it was supposed to be cheap to calibrate.? All you
> had to do to do a complete calibration was to unplug
> the module in the back panel, and replace it with a
> recently calibrated module.? The idea was you could
> order up a calibration by mail, and return your old
> calibration module as a core... never having to send
> the full meter in for calibration.
>
> Interesting idea, but surely not worth twice the price
> of a 3456A.
>
> I say it was a dinosaur, because it used all small
> scale logic rather than large highly capable hybrid
> circuitry, like the 3456A.? It should be much easier
> to repair, but it should also be much less reliable.
>
> And, that is my experience with the two units.
>
> -Chuck Harris
>
> paul swedberg wrote:
>> With respect to the 3455 and 3456 the difference is night and day. I have
>> numbers of 3456s and was given a seriously troubled 3455. Quite a
>> interesting meter.
>> Replaced at least 11 actually malfunctioning chips on the logic board. The
>> logics simple so easy to verify the chips were bad. No shot gunning
>> involved. Never really did get it working but ended up with a different
>> 3455 that needed help and the AC board and it is in fine shape. With
>> respect to the old 3455 it has both a logic and voltmeter issue. The 3455
>> uses an unusual processor and teh DVM uses the same proc and a masked eprom
>> that is known to go bad. It acts as a state machine.
>> The 3456s are losing there eproms and copies can be found on the net. Thank
>> heavens.
>> Regards
>> Paul
>> WB8TSL
>>
>>
>>
>>
>
>
>
>
>
>
>
>

You guys already know this, but just in case: an important consideration in rack mounting your gear is the weight that the floor can support.? With some of the taller racks filled with equipment, you will have hundreds of kgs. of weight.? As an extreme (and fabricated) example, a common 7 foot (44U) rack could physically accommodate eleven HP8664A signal generators.? At 35kg each, these generators would weigh a total of 385 kg. (850 lbs.) -- and this doesn't including the weight of the rack itself, and any mounting accessories.? Be safe, everyone!

On Apr 1, 2020, at 4:12 PM, Jared Cabot via <jaredcabot@...> wrote:

Ok, I've got everyone who has posted or messaged me on the list now.
I'll put the current list up this weekend so you can all double check and correct me if I made any mistakes.

?

A little bad news though.... Due to the Corona virus, Japan Post has basically shut down all deliveries to 153 countries as of yesterday....
It ultimately won't stop me from completing this, but it will delay things for a while, and I have no idea how long.

?

I'll keep the updates coming, and if I can get the PCB's delivered (The major carriers like Fedex, DHL, UPS, etc are currently still delivering into Japan) then pictures and a few performance tests will be forthcoming too.

Anyway, orders will remain open until the last minute, so keep the list growing! I got 53 PCB's to order on the list so far (and a big portion of those are full kits, so I'll have a good amount of parts to sort through too!)

It's up to you. How close do you want to be abl to set it?

A +-5V supply an 10 Turn pot is not too hard to do. Personlly I've used a isolated 12V output DC-DC converter with lots of filtering, a 10V reference IC a 1:1 resistive divider connected to the EFC common, and a 10k 10 tun pot.

Robert G8RPI

Some customers have been successful performing system recoveries with external CDROM drives from companies such as MicroSolutions. These drives connect to the scope’s parallel port. Since this method of recovery is unsupported by Agilent, customers are responsible for all configuration of the external drive. 8 For the recovery to be possible, you will need to mount the external drive and boot the scope from a floppy disk (the drive’s manufacturer may be able to provide this disk for you). At the DOS prompt, you must be able to navigate to the drive and see the drive’s contents with the dir command. If you can successfully get to this point, you should be able to perform a recovery with a special A.03.50 recovery CD. You may download the CD image at the location given below or contact Agilent technical support for a copy of this CD. The following recovery software is available for download. This is an .ISO CD-ROM image file. You will need to use CD burning software such as Nero or Easy CD Creator to create the CD from this file. Do not simply copy the file onto the CD as this will not work. A.03.50 Recovery CD: ftp://ftp.agilent.com/pub/callpub1/ddt/scope/548xx/Upg_Rec/3_5/3_5_rec.iso?