Well damn, now I'm really looking forward to it. I wish I wasn't so damn busy this week. I'm shipping prototypes of a new design to an absolutely neurotic client next week. :-/

-Dave

On May 7, 2024 8:15:02 PM "Dave Miller" <ve7hr@...> wrote:

I having been having fun with mine.
Not the same ambience as a 5370’s fan blasting away.
Documentation is somewhat cryptic.
Runs nice with Timelab.
Only hint is get the levels right for best performance.
Data looks good so far.
Noise floor is fantastic.

Dave
VE7HR

On May 7, 2024, at 4:56?PM, Dave McGuire via groups.io <mcguire@...> wrote:

?On 5/7/24 19:52, Ed (scskits) wrote:

Just got in a TinyPFA.
Hopefully I can get some more data on the GPSDO and other OCXOs that I have.
Any setups for the TinyPFA that can help with the learning curve would be appreciated.

I just got one as well, haven't even opened the box yet. I'm interested to hear about your activities.

-Dave

--
Dave McGuire, AK4HZ
New Kensington, PA

--
Dave McGuire, AK4HZ
New Kensington, PA

I am seeking some help with my HP 8566B which has a PL2 unlock error.
While I understand the use of the instrument, I need some guidance in fault finding.
The error occurs when the span is increased above 500KHz.?

The display then freezes and needs an instrument preset to work again.
The centre frequency does not seem make any difference.
I have not yet inspected the PL lock lights or checked voltages inside but I suspect they should be my first steps.

Any insight would be useful.

I have PDFs of the service manuals.

If there is anyone in Sydney, Australia, that would also be of help.

Thanks for the replies.

?

I am inclined to leave the resistors alone, but I just keep wondering how much their temperature coefficient and value drift with age would be affecting the accuracy of the instrument?

Against my Fluke 731B volt standard connected with Pomona low EMF cables, the first 3457A meter I repaired exhibited a drift due to temperature from 10.00016V to 10.00020V between 23 to 27 degrees C.
The second 3457A should arrive today or tomorrow so I'll be able to test it after I have had it powered on for a day or two to settle.
On my 34461A under the same test conditions, the voltage range was 9.99995V to 9.99996V


Jared

Yes Chuck, I know how EPROMs and EPROM programmers work. I've been using them as long as you have. I think we're talking past each other. There's something up with the 3852A EPROM images, as you probably saw in the rest of this rather muddy thread.

We need to ascertain what the actual EPROM types are, and ensure that we have images of them that are complete. Maciej's results with 4.22 point to an issue with that release, and HP's upgrade document doesn't do a great job of describing the nature of the upgrade and doesn't mention (at least not that I noticed) the sizes of the EPROMs.

-Dave

--
Dave McGuire, AK4HZ
New Kensington, PA

I'm not suggesting that at all.

If you read a 27256 in an EPROM programmer set for 27512, you
will get two copies of the 27256, one in each half of the 27512.

I'm suggesting that an EPROM usually maps 1:1 in the data space
of the processor, so if you have the following 27256's:

[1][2][3][4]

They will appear in a 27512 ROM set as:

[1 2][3 4]

If you read the 27256's as 27512's, you will see:

[1 1][2 2][3 3][4 4]

And, if there were really 2 - 27256's, and 1 - 27512,
and, not knowing any better, you read them all as 27512's:

[1 1][2 2][3 4]

Checksums can make a few boundary bytes different.

-Chuck Harris

On Wed, 8 May 2024 12:06:34 -0400 "Dave McGuire"
<mcguire@...> wrote:

Are you suggesting that HP changed from 27256 to 27512, just
putting two copies of the code in the low and high halves? I could
believe that, if 27512s were easier to come by, either in the
industry or within HP's purchasing power.

Maciej's most recent post indicated that he'd found that the
contents of U57 and U58 are not repeated into the high half, but it
was in the other six EPROMs. This is in 4.21.

So unless they changed this from 4.21 -> 4.22, the high halves of
4.22 for U57 and U58 are missing from the images taken by Anders and
posted on his 3852A page. I've asked him for clarification and am
waiting to hear back. He is on this list, though he hasn't been very
active here.

-Dave

On 5/8/24 11:56, Chuck Harris wrote:

Generally, changing from one EPROM size
to another doesn't change the code, but rather
just changes which EPROM it is put in.

I would take the images from the 32Kb set, and
convert them to binary images, and then do some
side by side comparisons between the 64Kb binary
images and the seemingly appropriate 32kb images
looking for similarity.

I expect that you will find that if you combine
the 32kb images in appropriate pairs, they will
match (mostly) the 64kb images.

-Chuck Harris


On Wed, 8 May 2024 10:16:24 -0400 "Dave McGuire"
<mcguire@...> wrote:

On 5/8/24 04:45, Dave_G0WBX via groups.io wrote:

You can of course put a 32k blob, in the upper half of a 27x512,
and it will work.? Put the same blob in both halves of a 64k EPROM
and you're golden.

Done all that soooo many times with other kit that had firmware in
32k images, but only 64k chips being available.

Of course. But from the upgrade document, and what Maciej and
Rick have said, it sounds like HP changed U57 and U58 from 27256 to
27512, involving a field PCB modification; one presumes they did
this for a reason. So what was supposed to be in the upper 32KB
of that 27512?

This whole thing is looking weird and we need some real info.

- All of the EPROMs in the 4.21 image set are 64KB.
- All of the EPROMs in the 4.22 set are 32KB.
- The upgrade document describes a modification involving the
high-order address bit for (at least) U57 and U58 for old boards.
- Maciej and Rick found that U57 and U58 need to be 27512s.

So as you can see, this situation makes little sense. Anders,
the author of the excellent "The HP3852A Page", got his 4.22
EPROMs from Keysight. I've asked him for clarification and to
verify the EPROM capacities for the 4.22 set.

From the upgrade document, it looks like they added jumpers for
27256 vs 27512 EPROMs from rev B to rev C of the processor board.

-Dave

开云体育

On Wed, May 8, 2024 at 03:27 PM, Bepi wrote:

I found the main power supply issue, a blown capacitor in the HV section, one side to ground. Unplugging the tubes was not enough to exclude the HV section.
From the schematic?

I notice that the blown capacitor C112 has a sister one C111 , same value, according to the component list 0.1 uF

if I measure the good one i get a 3 times bigger value? (this Chinese res/cap/ind/diode/transistor 3$ is amazingly useful)

?and the resistor in between measures 45 ohm instead of 100. Can anybody read the capacitor color codes?
Reality is different from the schematic , the manual is for mod 25 to 724, mine is N 2482
I am tempted to go with the schematic and replace the original capacitors with polyester ones I already purchased, .1uF 650V, am I doing the right thing?

Following the manual the 300V rail is regarded as part of the low voltage power supply.

Unplugging the tubes for the EHT should prevent damage to the unobtainium EHT transformer, while you fixing the rest. Of course you should find/replace any paper capacitors on the EHT rails before thinking about plugging them back in, there are several 5kV capacitors listed as "paper" based, these need to be replaced if they haven't already been done in the past.

C111 & C112 are 0.1uF for all serial numbers.

The 3$ tester is no good for checking these capacitors, something that tests at the working voltage is much better for identifying troublesome paper or mica capacitors in valve/tube equipment, try & buy/obtain/borrow an insulation tester & learn how to use it.

David

Very nice work indeed! Thanks for making such careful measurements. That data is pure gold.

-- Cheers
Tom

--
Prof. Thomas H. Lee
Allen Ctr., Rm. 205
420 Via Palou Mall
Stanford University
Stanford, CA 94305-4070

On 5/8/24 13:31, gren wrote:

Julio's last msg to me was:
"? I purchased a $60 programmer on Ebay made in China, and used it to duplicate the ROMS,
??? However, the included software was tricky to use, and the 27C512 devices could NOT be program
??? under Windows 7 64 bit. U57 & U58 had to be programmed using an old computer I had with Windows XP,
??? it worked!!? "

Where Julio bought his programmer, where it was built, which one it was, or what OS he was using to drive it has no bearing whatsoever on the size of the EPROMs or the correctness or lack thereof of the 3852A firmware images that are floating around. Once again, this is just muddying the waters.

-Dave

--
Dave McGuire, AK4HZ
New Kensington, PA

开云体育

Dear Yves,

None of which has anything at all to do with the contents of the EPROMs, their sizes, or the PCB modification for the high-order address bit!

The manufacturer and device ID that can be read out in ID mode is the sole responsibility of the programmer to deal with or ignore. Julio's cheap Chinese EPROM programmer's inability to deal with a standard component has no bearing whatsoever on this whole situation.

-Dave

--
Dave McGuire, AK4HZ
New Kensington, PA

I hope this isn't too much info, but see below for VNA derived plots of the parallel Rp and Cp of the HP 8405A diode when reverse biased by 3V. It's also compared to the 1N6263 diode across 10-100MHz.

In reality, the Cp will be about 0.2pF lower than this because my test fixture has about 0.06pF (uncorrected) fringing capacitance and I had to test the diode without cropping the legs any shorter because I had to refit them back into the probe. So there could easily be another 0.15pF capacitance to free space in the test fixture. The main thing to do is look at the difference. 0.48pF for the HP 8405A diode vs 1.25pF for the 1N6263A diode at 3V reverse bias.

The other plot shows the small signal dynamic resistance at 10mA forward bias. The earlier curve trace plot predicted about 11 ohms for the HP 8405A diode and the VNA plot below confirms this. It also confirms that the dynamic resistance of the 1N6263 diode is much higher at about 40 ohms. So this is another clear difference between these two diodes.

I haven't bothered to put the NXP diode or the HSMS-282Z on the curve tracer or on the VNA as these diodes aren't close enough to the HP 8405A diode. However, I can do this at some point in the next few days if this would be of any interest.

The BAT-62 looks like it could be a good substitute diode but I won't know until I get one delivered. I will definitely put this diode on the curve tracer and the VNA. I'm hoping it will be the best diode yet...

All fixed. Cost more in postage than the parts required!

The source chain consists (after the mixer) of 2xERA3s 1xERA4, a small tranny and a big bolt down jobby, thankfully the latter 2 were fine and just the 2nd ERA3 and the ERA4 were Kaput. No idea why this'd happened. They were still taking reasonable current. Very easy debug and fix and put that down to the excellent build quality and construction. A joy to work on!

There's no need to change those 51 k resistors. HP had it all figured out, for proper and safe input overload handling and ESD. You'll probably notice they are fairly large (maybe 1 or 2W) in power rating. That comes into play in the event of sustained over-voltage on the input connection, where the voltage clamps in the front-end limit the circuit voltage, and the resistors limit the current and take the rest of the total voltage, and have to dissipate the power. To appreciate the size needed, just say, take the specified maximum input overload voltage squared over the total R, for total power. Normally, these parts seldom need to do their thing, but they're always ready. If they appear overheated or burnt, then the unit may have been subject to severe abuse.

Ed

Are you suggesting that HP changed from 27256 to 27512, just putting two copies of the code in the low and high halves? I could believe that, if 27512s were easier to come by, either in the industry or within HP's purchasing power.

Maciej's most recent post indicated that he'd found that the contents of U57 and U58 are not repeated into the high half, but it was in the other six EPROMs. This is in 4.21.

So unless they changed this from 4.21 -> 4.22, the high halves of 4.22 for U57 and U58 are missing from the images taken by Anders and posted on his 3852A page. I've asked him for clarification and am waiting to hear back. He is on this list, though he hasn't been very active here.

-Dave

--
Dave McGuire, AK4HZ
New Kensington, PA

Both U57 and U58 have valid data up to address CD18 (52504) . So what ever is in the ROM almost uses the 80% of the 65535byes available.? Definitely this will not fit on a 32K chip .? There must be a SW mechanism which we saw in action when I burned 4.22? that gives errors about the modification not being complete. The SW is most likely checking if it can reach the code passed the first 32K.? If it cant it fails and responds with the "ROM update not complete " message.? Very good engineering on HP part.?

Here is U58 at the point where valid data ends.?

Generally, changing from one EPROM size
to another doesn't change the code, but rather
just changes which EPROM it is put in.

I would take the images from the 32Kb set, and
convert them to binary images, and then do some
side by side comparisons between the 64Kb binary
images and the seemingly appropriate 32kb images
looking for similarity.

I expect that you will find that if you combine
the 32kb images in appropriate pairs, they will
match (mostly) the 64kb images.

-Chuck Harris


On Wed, 8 May 2024 10:16:24 -0400 "Dave McGuire"
<mcguire@...> wrote:

On 5/8/24 04:45, Dave_G0WBX via groups.io wrote:

You can of course put a 32k blob, in the upper half of a 27x512,
and it will work.? Put the same blob in both halves of a 64k EPROM
and you're golden.

Done all that soooo many times with other kit that had firmware in
32k images, but only 64k chips being available.

Of course. But from the upgrade document, and what Maciej and
Rick have said, it sounds like HP changed U57 and U58 from 27256 to
27512, involving a field PCB modification; one presumes they did this
for a reason. So what was supposed to be in the upper 32KB of that
27512?

This whole thing is looking weird and we need some real info.

- All of the EPROMs in the 4.21 image set are 64KB.
- All of the EPROMs in the 4.22 set are 32KB.
- The upgrade document describes a modification involving the
high-order address bit for (at least) U57 and U58 for old boards.
- Maciej and Rick found that U57 and U58 need to be 27512s.

So as you can see, this situation makes little sense. Anders, the
author of the excellent "The HP3852A Page", got his 4.22 EPROMs from
Keysight. I've asked him for clarification and to verify the EPROM
capacities for the 4.22 set.

From the upgrade document, it looks like they added jumpers for
27256 vs 27512 EPROMs from rev B to rev C of the processor board.

-Dave