EEVblog #523 - REPAIR: HP 35660A Dynamic Signal Analyser

[nitro2k01]

Oh, I forgot to mention this. When you put a metallic screwdriver in the width adjust hole, it disturbed the adjustment, as expected. But I noticed that this made the image more narrow. You could possibly use this to your advantage. First adjust the coil to produce the minimum width, then permanently put some long metallic object in the hole, which will narrow down the image a small bit more. Then fix this object to the case with some semi-permanent method, such as a small dab of hot snot.

[SeanB]

If you want a tube I can post you one, complete in the box with the rest of the monitor. Orange and not green, but it still works well, and you just transplant it complete with the yoke, it will either plug straight in or just match the colours. Might have a few slight screen burns on it but otherwise working. Going to be expensive to airmail, but sea will take about 8 weeks to do it. I had more but had a clean up last month and tossed all the dozen or so CGA monitors that were still around. Still have one EGA one left up on a shelf though

[G7PSK]

I think that to pull the picture back to shape you need a key to rotate each of those magnets in the cage around the CRT.
They are there top counter balance the earths and other magnetic fields. I can remember fiddling with them in the past you turn one and the picture distorts one way then you have to turn another to pull that and so on until the picture comes right. A real PTA. That unit had quite a shock to knock the gun connector board off as well as those PM's probably moved the yoke back as well on the other hand it could have been subjected to high levels of vibration over a period of time such as in an aircraft or truck.

[PChi]

I always thought that Hewlett Packkard had let the metal work quality slip with the HP35660A. When I picked up the unit to move it my fingers didn't like the sharp corners on the box. I always assumed that it was primarily designed to be rack mounted.
I'd say the box is of some use because (if I remember correctly) it goes down to near DC (input can be DC coupled) that a PC sound card wont do. Also the absolute levels are known without requiring a calibration oscillator and procedure.
It also has a 16 bit A/D so better dynamic range than an Oscilloscpe and possibly better than actual PC performance.
From memory I was under the impression that the noise performance of the input amplifier was nothing special.

[Ketturi]

Dave you poked CRT with plastic black (ESD one I presume) adjustment tool! Last time I did that I got pretty nasty bite from high voltage discharge.

[EEVblog]

From memory I was under the impression that the noise performance of the input amplifier was nothing special.

No, it's not. You aren't going to be measuring leading 1/f noise with it.

[Salas]

From memory I was under the impression that the noise performance of the input amplifier was nothing special.

Just looked in it. It is quoted as -130dBV/SQRT(HZ) from 160Hz to 1.28kHz and -140 from 1.28-102.4k.
That will translate to a 5k5 Ohm en in the first case. I.e. no better than 9.54 nVrtHz performance in that band.
Surely no good to characterize ultra low noise semis with low frequency 1/f corner with this HP DSA model.
Still good for checking non ULNA preamps, average NF quoted BJTs and JFETs, Zeners, chip regulators, etc.

[EEVblog]

Just looked in it. It is quoted as -130dBV/SQRT(HZ) from 160Hz to 1.28kHz and -140 from 1.28-102.4k.
That will translate to a 5k5 Ohm en in the first case. I.e. no better than 9.54 nVrtHz performance in that band.
Surely no good to characterize ultra low noise semis with low frequency 1/f corner with this HP DSA model.
Still good for checking non ULNA preamps, average NF quoted BJTs and JFETs, Zeners, chip regulators, etc.

Yep, not the best, but still good for tons of uses.
I was just thinking that it might be possible to upgrade it with modern really schmick opamps. The service manual does not have a BOM, so no idea what opamps are used.
I presume it would be mostly driven by the opamps, and the FET input stage is unlikely to dominate.

[Psi]

It is possible that the unit was mounted in a rack. The rack may have been accidentally dropped on its back

That's the best explanation so far.
It would provide enough force to move the yoke and knock the board off while leaving the case undamaged.

[EEVblog]

BTW, I can confirm it was the yoke moved slightly back on the neck, I wasn't rough enough the first time.
Followup video has been uploaded and will be released tomorrow.
Image is now just fine.

[Salas]

Just looked in it. It is quoted as -130dBV/SQRT(HZ) from 160Hz to 1.28kHz and -140 from 1.28-102.4k.
That will translate to a 5k5 Ohm en in the first case. I.e. no better than 9.54 nVrtHz performance in that band.
Surely no good to characterize ultra low noise semis with low frequency 1/f corner with this HP DSA model.
Still good for checking non ULNA preamps, average NF quoted BJTs and JFETs, Zeners, chip regulators, etc.

Yep, not the best, but still good for tons of uses.
I was just thinking that it might be possible to upgrade it with modern really schmick opamps. The service manual does not have a BOM, so no idea what opamps are used.
I presume it would be mostly driven by the opamps, and the FET input stage is unlikely to dominate.

The head amp will dominate the SNR as long as its gain isn't less than 1/3 the total closed loop gain of the input preamp's stages.
Surely worth looking at its gain structure in the service manual and decide where the upgrade will likely be most effective.

[EEVblog]

Surely worth looking at its gain structure in the service manual and decide where the upgrade will likely be most effective.

Yeah, need to check. IIRC the FET input stage is just a buffer.

[Salas]

Surely worth looking at its gain structure in the service manual and decide where the upgrade will likely be most effective.

Yeah, need to check. IIRC the FET input stage is just a buffer.

They combine a FET source follower with a bipolar input op-amp just for high input impedance then?
In that case that op-amp's noise spec will dominate the whole preamp's spec as long as it does a good portion of the total gain.

There are enough low noise op-amps these days to choose. The AD797 can hit 1nVrtHz at 100Hz for instance.

[bktemp]

It is possible that the unit was mounted in a rack. The rack may have been accidentally dropped on its back

That's the best explanation so far.
It would provide enough force to move the yoke and knock the board off while leaving the case undamaged.

If the device really got a hard hit, it would be a good idea to check the power supply. Large capacitors and the transformer are heavy parts and can easily be ripped of the board like in this device:

Even if it still works, there could be bad solder connections which will lead to arcing and slowly destroy semiconductors. I have seen such failures a few times in TV sets an PC monitors around the heavy flyback transformers where broken solder joints destroyed the horizontal output transistor due to voltage spikes.

[envisionelec]

I maintain that it was a sustained vibration incident rather than a blow to the chassis. The metalwork would be damaged internally which would be evidenced by difficulty of access such as sliding the monitor out or removing the lid.

[RupertGo]

BTW, I can confirm it was the yoke moved slightly back on the neck, I wasn't rough enough the first time.
Followup video has been uploaded and will be released tomorrow.
Image is now just fine.

Ah, fabulous. One of the occasions it's really good to be proved wrong.

[NiHaoMike]

Might it be possible to replace the CRT module with a composite LCD?

[M0BSW]

I enjoyed both video's repair and explanation 

[free_electron]

I am working on replacing the tube with an lcd.
There is a company that offers replacements. The monitor in the analyser (matsushita) was very popular in cnc machines. There is a someone that sells a 400$ lcd monitor that is compatible with this matsushita one.

I am trying to get an lcd panel to run.
I have a panel from an infiniium scope. This is bstandard tft panel (not with lvds but with normal inputs)
Basically the lcd panel has 24 bit for color. A hsync , vsync and a clock.
Driving such a panel is easy. Assert both hsync and vsync to set the picel cursor top left. Apply data on the rgb bus and give a clock pulse. When you reach end of the first line you give a hsync to move to the next line. (Hsync without vsync means move to next line. Is both are asserted it means reset to top left)

I will wire the rgb data to fixed values and use the msb of r,g and b to select full or half brightness.

I think a small cpld could do the work. Hsync and vsync come out of the graphics card of the machine.
The itemsity bit and blankin also , so those will drive the rgb.
A local clock oscillator in the cpld will provide the pixel clock. I need to set the speed of that oscillator in such a way that it provides avout 720 pulses between hsyncs. ( these screens are mda 720x400 i believe)

The panels i have are 800x600 so it should work perfectly.
It'll take me some time to build it. I'll post results.

[EEVblog]

They combine a FET source follower with a bipolar input op-amp just for high input impedance then?
In that case that op-amp's noise spec will dominate the whole preamp's spec as long as it does a good portion of the total gain.

Yep, looks like it.
I need to get back in there and see what parts are used.
Would also be awesome if the opamp was socketed, but of course it won't be.

[Salas]

They combine a FET source follower with a bipolar input op-amp just for high input impedance then?
In that case that op-amp's noise spec will dominate the whole preamp's spec as long as it does a good portion of the total gain.

Yep, looks like it.
I need to get back in there and see what parts are used.
Would also be awesome if the opamp was socketed, but of course it won't be.

Does it have the ADC resolution and Y scale options on the screen display to support a possible 100dB+ SNR spec upgrade in the analogue domain?

[mmagin]

For my amusement, I recently got one of these for "parts or not working", though it didn't power up nor show a display when I got it.  Things that needed repair:

• Did some CRT rejuvination (ala http://www.ke5fx.com/crt.html ) and turned up/adjusted the CRT's controls
• Had vignetting also!  Loosened the clamp and slid the yoke forward and tilted it to level, very gently tightened the clamp
• Replaced a front panel BNC that was dented up
• Since the particular BNC connector is no longer in production, I bought the whole front BNC board on ebay and replaced it.  Good thing too, one of the old MOVs was toasty, they mean what they say about floating only +/-4V to ground

I think the floppy may still be non-functional, and I haven't had a chance to check any of the calibration aside from running the self-test/self-cal stuff.  Too bad the only scan of the service manual out there is incomplete, and too bad I don't have much space for boatanchors :/

[Pezo]

Dave,
did you by any chance keep a copy of that service manual around that you could upload somewhere? The link doesn't work any more, and I could only find a broken copy of it that is incomplete and seems to have several other manuals mixed in...

Thanks, Peter