No Load on Backlight Inverter (HP 8904A Display Mod)

[wn1fju]

Back in 2015, Roberto EB4EQA posted a mod to the HP 8904A function generator to replace the display, with its very weak backlight, with a more modern 40x2 display (with LED backlight).  In that mod, he suggested simply unplugging the two wire connection that supplies nominally 80 VAC to the old display's backlight.

Makes sense to me, but then I came across this sentence in the HP 8904A service manual:  "A no load or short on the backlighting oscillator of A2 will damage the drive circuit."  Not that I am going to need the existing backlighting function, but I'd rather not destroy parts needlessly.

A2 is the main board of the HP 8904A and there is a 3-terminal device (DC in, AC out, common) in a little black box that acts as the inverter for the backlight.  I happen to have a partially destroyed spare A2 board (battery leakage) and pulled the inverter module out.  I gave it 3 VDC and it produced 80 VACrms.  With a
500K load across the output, it was drawing 17 mA on the input.  With no load, I still got 80 VAC, but obviously less input current.

So my questions are:  1) What are the perils of the no-load condition?  2) Is 500K a good choice for a load? 

[TheMG]

Why not just remove the module, or disconnect the input power from it? Seems like a better solution that dummy loading it and using unnecessary power.

[wn1fju]

Yes, removing the inverter module would certainly prevent it from damage, although it does require removing the entire board.  I was looking more for an explanation of why an unloaded inverter would be a problem.  Incidentally, the software does cut the power to the inverter after 3 minutes of no keyboard
activity, so not a lot of power would be used if I left it in there.  The new display's backlight will remain on all the time, thankfully.

Not that I use the 8904A much anyway.  In my opinion, it is not one of HP's better efforts.  First, they call it a DC-600kHz synthesizer, but that's only for sine
waves.  The triangle, ramp, pulse waveforms only go to 50 kHz.  And since it is wave-table based, many of the waveforms suffer from address rounding which
gives jittered results (there is a whole discussion of this in the manual).  Moreover, they priced the thing absurdly which required the purchaser to buy options which should have been included in the first place (fortunately it can be easily hacked via software to enable the options).  Then, of course, if the backup battery dies, you
lose the serial number/options and supposedly have to return the piece to HP (fortunately, that can be hacked/restored via software too).  And the final insult is the service manual which omits the schematics/parts lists for the main board with the "excuse" that it is an HP-proprietary design.  That's my mini-review of the HP 8904A.

[TheMG]

At 80VAC I'm assuming the original backlight was an EL (electroluminescent) type?

Since the voltage seems to remain in spec even without a load, it could be that the inverter design relies on the inherent capacitance of the EL backlight to filter out any switching spikes/transients that may otherwise damage the switching transistors in the inverter? Or maybe there's some components in there to clamp the maximum voltage that may over time overheat and fail if operated without a load for too long.

Who really knows. Looking at the waveform on a scope might shed some light on it.

[free_electron]

hack ? hacK ? how ? i have one of these things .

[wn1fju]

I looked at the inverter output on my working HP 8904A.  It was a slightly distorted sine wave, 80 VACrms @ 523 Hz. 

I then played around with the inverter I pulled from a non-working 8904A board.  Incidentally, it is a different shape, size, color and orientation than the other one, so I guess HP at one time redesigned the board.  With close to 3 VDC input, I got 80 VACrms @ 2.46 kHz open load and 74 VACrms @ 1.98 kHz with a 500 Kohm load.  The waveforms were a bit more distorted than with the inverter hooked up to the actual display, but not too bad.  Nothing that I would call a "spike."

The main board doesn't look too bad to remove, so I guess I will simply remove the inverter module.

To free_electron who said "hack? hack? how?":  Are you asking how to enable the options and/or restore the serial number?  This can all be done by entering commands from the front panel.  Search "HP8904A Options" on the eevBlog forums.  I did it years ago and it is a simple recipe.

[TheMG]

Maybe at one point once upon a time they had a version where it actually did matter. That's the only thing I can really think of. If the voltage stays in tolerance and the waveform doesn't turn ugly with no load then I don't really see any reason it would be damaged by running without a load, at least with the version of inverters that you have. Just speculation at this point really.

[DeepLink]

Nice to see that some 8904A are still in use - so is mine from time to time

Not to hack this tread, but when you have your 8904A open, please check your firmware version
I, and at least one other, are in need of a newer version (08904-87010)

Check: https://www.eevblog.com/forum/testgear/hp-8904a-firmware-eprom-image-08904-87010/msg2952304/#msg2952304

[wn1fju]

To DeepLink:  I opened my 8904A up (S/N 3633Axxxx) and took a look.  The EPROM is marked 87011.  Sorry that doesn't help.

[DeepLink]

wn1fju - Thank you for looking

Mine is a 2822A series, so a bit older

[texaspyro]

The EL backlight inverter in Tek 1502A amd 1503A TDRs will definintely fry if operated without a load...