Tnx Chris!

I expected your reply and it was very useful! I know that remote-troubleshooting is impossible, but your advises (e.g. by starting from Section V, Vol. 1) seem to be golden. By the way I should like to share - when strictly needed - some photo/video of my 3585A, with you and the other interested people (obviously I will post a link for downloading).

Alberto, IZ2EWV

Hello Alberto,

having been down the 3585A rabbit hole twice, I can tell you that remote-troubleshooting this instrument is very difficult. A lot of effects and defects can cause the "LOCAL OSC. UNLOCKED" message, not only problems on A34. Also, the LED messages can mislead about the origin of the defect. On page 6-25/6-26 of Vol.1 of the Service Manual, you can see which other PCAs are part of the LO circuitry. As I recommended, please start with a full adjustment of the instrument (Vol. 1, Section V). You will reach a point where the adjustment will fail. This is your starting point for troubleshooting. The engineers at HP willingly/unwillingly set up this chapter in a way that it helps to find exactly the LO unlocked-related problems (if the problem is a singular problem. Not guaranteed, keep your fingers crossed!).

I bought both of my 3585As as defective units and both showed similar problems as yours. First unit: at startup "CAL ERROR 02/07" message, wrong LED status on A34, then diverting into "LOCAL OSC. UNLOCKED" message, beeping every other second. It also showed an elevated noise level, along with a reduced sensitivity. It turned out that the reduced sensitivity came from a blown first mixer (CR2 on A4), the elevated noise level (and the CAL ERROR messages) came from a defective mixer/amplifier (U1 on A5), the LO unlock message finally was fixed by repairs on A33 (defective divider, failed when warming up) and an underperforming OP amp on A25. Because I assumed multiple failures as the reason (and being very disappointed about the lack of success), I bought a parts donor unit to at least be able to localize the affected PCA. It was a 'flat liner' as yours, beeping and blinking in the same way and showing the infamous LO unlocked message. Turned out that this was not really helpful for pinning down the problem, but I could at least sort out some effects. On the way, I realized that a parts donor might be a parts donor for a good reason, and before I started to swap PCAs with my first patient, I diverted into checking which parts of this unit are supposedly defective. Finally, the second unit had a defective main oscillator A81 (ultra stable, but happily swinging at about 11.2MHz, not sure how that is possible), an A52 Tracking Generator with no output (that did not help because you need it for service tasks), a broken J2 SMB connector on A21 (quite obviously, somebody was in there and punched the card back in place with a sledgehammer), a noisy log amp (U5) on A14 and a defective summing amp on A27.? The calibration was completely out of whack, it seemed like the previous owner tried to squeeze out whatever is possible, but finally did not succeed because of the defect (s).
These repairs were just a nightmare, but I earned a lot! The first unit is sitting aside my workbench and as the second unit needed a job, I finally decided to build a 3047A system with it (3582A/3585A/35601A/9836A)...

Chris

On 6/6/21 11:02 PM, Matt Huszagh wrote:

When I receive a new piece of vintage equipment I like to back up the
firmware in case the EPROM containing it one day becomes
damaged. Equipment often also contains calibration constants in a
separate EPROM.

What are these calibration constants? How does the device firmware use
them at a high level? Or, if that question is too broad, what are some
examples of how that data is used?

Calibration constants are numbers that are generated during a
calibration procedure, starting at the factory and possibly regenerated
during later recalibration. These numbers are constants that describe,
essentially, the difference in behavior of an instrument or assembly
from "ideal", and are used as correction factors. Different instruments
or assemblies will have different types of calibration constants
depending on their purpose.

For example, you've seen sensor heads for HP RF power meters that have
a list of correction factors on the side. These numbers describe the
measured behavior of that sensor at the listed frequencies. For the
mid-aged power meters like the 437B and 438A, you enter those numbers
into the meter, tell the meter what frequency you're measuring, and it
applies, those correction factors to the displayed data.

Later HP RF power sensor heads have those correction factors stored in
an NVRAM chip within the sensor head, and they're read out by the power
meter at power-up time. This saves the operator from having to enter
those numbers, but if it's lost, it's very difficult to regenerate.

Are these calibration constants particular to a specific unit? For
example, would 2 8340B's have the same calibration constant data?

No, these are per-instrument. (or per-assembly, or per-component)

Is there any reason why I might not want to, or it might not be
necessary, to backup this data? For firmware, I can often track down
copies of the firmware online (e.g., at ko4bb.com). I have not found
this to be true for the calibration constants. Is this because it is
unit-specific and therefore would not make sense to share?

Yes, they are unit-specific. Calibration constants from a different
instrument of the same model will usually be useless. If your data is
completely lost, constants from another identical model may get you
somewhat close, perhaps close enough to run the instrument, but there
will be inaccuracies. Some types of instruments have calibration
constants that are so critical to the internal operation of the
instrument that the instrument will fail power-up self-tests, and even
fail to function at all.

You should try hard to not lose any calibration constants from any
instrument. In some cases they can be regenerated, but in many such
cases the procedure is difficult and requires very specific equipment to
perform.

-Dave

--
Dave McGuire, AK4HZ
New Kensington, PA

When I receive a new piece of vintage equipment I like to back up the
firmware in case the EPROM containing it one day becomes
damaged. Equipment often also contains calibration constants in a
separate EPROM.

What are these calibration constants? How does the device firmware use
them at a high level? Or, if that question is too broad, what are some
examples of how that data is used?

Are these calibration constants particular to a specific unit? For
example, would 2 8340B's have the same calibration constant data?

Is there any reason why I might not want to, or it might not be
necessary, to backup this data? For firmware, I can often track down
copies of the firmware online (e.g., at ko4bb.com). I have not found
this to be true for the calibration constants. Is this because it is
unit-specific and therefore would not make sense to share?

Thanks
Matt

Rick

I checked out the device you have mentioned, it appears to be a glorified bread board with just about everything built in which I think is quite neat, thanks for sharing!

On June 6, 2021 4:10:04 PM "Rick - WA6NDR" <nungester@...> wrote:

In all function generators I know about, both square waves and triangle waves have a default DC Offset of 0 Volts. (They go above and below 0V equally.) Adding in a DC offset is usually a separate setting.

Not HP equipment but related: I recently gave a talk to a local ham radio club about 'scopes and spectrum analyzers using a Digilent Electronics Explorer board ( ) and its Waveforms software ( ). It is a great "electronics lab on a board" for experimenting and demonstrations. The following screenshots are from the talk, showing an audio spectrum analyzer image of a 1 kHz square wave and how that square wave looks on a scope after only 4 component sine waves are added together. (During the talk these are live images that update real-time as settings are changed.)

That last one is a very nice rendition of the summation of a Fourier series. It was a diagram very much like that one that got me over the hump to understand how they work.

-Dave

--
Dave McGuire, AK4HZ
New Kensington, PA

Also, common FGs are capable of putting out quite a bit of power, so it's good to be wary of the actual level - accidentally setting the FG's output improperly could put the SA at risk. A DC block, and a simple limiter (easy to put together for typical FG frequency range capability), or plenty of fixed attenuation right at the front of the SA would help keep things safe against operator error.

Ed

Mark

You are correct, I found the 9060 ohm SMD resistor had changed value
Replaced, now good, had to cobble a couple resistors together, but they fit in the case.

Thanks?

Wally KC9INK

In all function generators I know about, both square waves and triangle waves have a default DC Offset of 0 Volts. (They go above and below 0V equally.) Adding in a DC offset is usually a separate setting.

Not HP equipment but related: I recently gave a talk to a local ham radio club about 'scopes and spectrum analyzers using a and its . It is a great "electronics lab on a board" for experimenting and demonstrations. The following screenshots are from the talk, showing an audio spectrum analyzer image of a 1 kHz square wave and how that square wave looks on a scope after only 4 component sine waves are added together. (During the talk these are live images that update real-time as settings are changed.)

开云体育

Do you have a source for it being MFM (Double Density)? Everything I can find, and the age of the instrument, indicates FM (Single Density).

Indeed, it's FM, single density => my fault, still under impression of making images for 4145b.

IMHO, it's not that difficult to build the cable to the PC disk controller. Then the 4145a FDD can be removed and temporarily connected to the PC-compatible disk controller, which supports 256 bytes/sector, for use with ImageDisk.


BTW, does anyone from 4145b camp have a disk with the 4145b sample programs (originally supplied with the system diskette)?

This is totally doable. If you're worried about DC, which is good practice, a DC block will take care of it for you. These are cheap and widely available. E.g.?
$15

On the other hand as a measurement stimulus the usefulness is somewhat questionable. At least for cable or other 2 port component characterization. If your fundamental frequency is big you risk missing suckouts or ripple in between the squarewave harmonic peaks. If the fundamental frequency is low then the rolloff will be very quick. The 7th harmonic of a square wave only has about 1% of the signal power (-20dB). E.g. a 1MHz -30dBm square has -50dBm power at 7MHz.

Has anyone used GPIB to capture screens on this unit? If so, get me started in the right direction.

开云体育

Hello friends,

I'm repairing my hp 3585A, which shows "LOCAL OSC. UNLOCKED" on the screen all the time and a flat horizontal plot at the bottom of the graticule. I'm following the service manual and its troubleshooting trees and signatures, and I found a first fault (missing +5V) in the A34 board, which I repaired by replacing a 2N3055.

Now the voltage is OK and the A34 circuits receive all the voltages within the requested tolerances, but the 4 LEDs don't behave the way that they are supposed to, at least according to the "section 11, service group B" (recalled by the service manual if this message appears):

• the REF LED should be OFF and it is true for the first minutes of operation, then it lights a stay ON;
• the FRN LED blinks slowly (ON for about 5 s and OFF for 0,5 s) and every time it lights the sounder makes a beep.

I checked the A34 board almost entirely, measured frequencies, voltages and waveforms at the requested SMB, TPs and pins; I also replaced an IC (U8) that the signature analysis told me to be "possibly" defective. But it wasn't.

I'm unsure about the correct operation of the trigger, because its troubleshooting tree expects to see "pulses" at TP2, but I see only long changes of state whenever I put the Ext. Trigger input to ground, and even in the Line Trigger mode (high for 15 s, low for 2 s).

If anybody of you had experience in repairing this beautiful instrument, I as him some help, mostly for interpreting correctly the service manual and the signals that the hp 3585A gives with the timing of its LEDs. I usually restore, repair and align my radios and electronic instruments, but in this case I feel that I may misunderstand.

Tnx in advance,
Alberto, IZ2EWV

I received this email that I thought might interest other RF junkies:

Demo: PNA-X Spurious Test

Dr. Joel Dunsmore will walk through a 20-minute demo showing optimized spurious testing with the enhanced PNA/PNA-X new DDS source.

In this webinar series Dr. Dunsmore will demonstrate:
??? Close-in carrier spurious
??? Higher-order mixing products
??? Spurious vs. LO drive level
Join this Keysight webinar to learn how to optimize spurious tests on the PNA/PNA-X.


Date: June 16, 2021


https://connectlp.keysight.com/SpuriousTest-Americas?elq_cid=1130946&cmpid=ELQ-14600&elqCampaignId=14600&elqTrackId=2FE217D7C4FF72BD38DC574DD1E4E6B9&elq=ba1aa6467a24496da8e3fb5eff8200a8&elqaid=30097&elqat=1&elqCampaignId=14600

Great replies, thanks

Hi Wally,

There is a resistor inside the probe's compensation box that informs the scope of the probe's multiplication factor.? Check the resistance from the 1160A's AutoProbe pin to ground and compare it to your N2843A.? Maybe the resistor has gone bad.

-mark

On Sun, Jun 6, 2021 at 06:50 AM, <maxim.vlasov@...> wrote:

Hello,

Just a few notes about the 4145a format. There is no magic about it and it's a standard MFM (not GCR format fortunately).
Since 4145a as well as 4145b both use Fujitsu controller MB8866, which is a +5V only version of the famous 1793 (FD1793, FDC1793, SAB1793,...) integrated in many many home and office computers in the 80s.
In other words, even on Sinclair ZX-Spectrum (TRS80, Atari ST, MSX etc) a motivated coder can write a disk copy utility to duplicate 4145a floppy disk w/o any problem. It's not like Amiga floppy, which is impossible to copy on any PC FDC, not even by reading the whole track at once. So, it seems akward, but there is a better chance to use an old computer (eating dust in the attic) then the modern one.

"So a PC with a TEAC FD55A or B and one of the universal copy programs (dos or linux) should work."
Not every integrated PC FDC is 256 bytes/sector compatible (this is what's tested by ImageDisk utility), but for any FDC compatible with 256 bytes/sector copying wouldn't be the problem - sure.

Do you have a souce for it being MFM (Double Density)? Everything I can find, and the age of the instrument, indicates FM (Single Density).

Robert G8RPI.