Agilent 33250a Blown Output Hybrid

[daveyk]

Anybody ever able to fix or replace the output hybrid? Agilent will not sell the part; they swap generators for about $3500.

I would love to have someone design A discrete amplifier to replace the hybrid. I would only worry about 100KHz - 30MHz. The output amplitude doesn't have to remain flat as it would be monitored with a scope. It needs to drive 10vP-P in to 50 ohms. It would be great if we could fit a little add-on board inside.

Dave


Sent from my iPhone using Tapatalk

[TheSteve]

I've never seen anyone try to replace/repair the hybrid. I know I was warned to watch out for dead hybrids when looking to buy a 33250A to repair as it was game over if it was dead.

[daveyk]

I've never seen anyone try to replace/repair the hybrid. I know I was warned to watch out for dead hybrids when looking to buy a 33250A to repair as it was game over if it was dead.

Well this was from dumpster diving at work. The sine wave coming out of the attenuator network is pure.

The output must have been hit with a UT square wave pulser. The protection diode is also shorted. The diodes for the output monitoring circuits have been pulled so that it's possible to work on the generator. It thinks it's working. The hybrid is not. What I want is a replacement output amplifier. It only needs a 30MHz bandwidth, not 80MHz.

Maybe I should look to the schematic of an old Wavetek 166 for inspiration but was hoping an engineer on here may have already tackled this.

Dave


Sent from my iPhone using Tapatalk

[TheSteve]

One option is to sell it for parts on ebay - even with an advertised blown output you'll probably still get enough to buy a nice working 20-30 MHz AWG.

[james_s]

I really wish companies that make this sort of stuff would product a bunch of spare hybrids. In general I really hate hybrids and custom silicon, at least the latter is usually fairly reliable, hybrids seem to be a lot more problematic.

[HighVoltage]

The 33250A sells new for over $ 6000 !
I agree with TheSteve, sell it as broken and you might get as much money as you need to buy a working 30 MHz 33220A for around $500

Can you show us a picture of the broken hybrid?
I don't have a 33250A, So I have never seen inside one.

[daveyk]

The 33250A sells new for over $ 6000 !
I agree with TheSteve, sell it as broken and you might get as much money as you need to buy a working 30 MHz 33220A for around $500

Can you show us a picture of the broken hybrid?
I don't have a 33250A, So I have never seen inside one.

When I get home. It's probably 1"x2" with a heat sink bonded to it.

I wrote software around the 33250a, but if the 33220a had glib I could adapt the test program. I'll look in to it; thanks.

Dave


Sent from my iPhone using Tapatalk

[daveyk]

The 33250A sells new for over $ 6000 !
I agree with TheSteve, sell it as broken and you might get as much money as you need to buy a working 30 MHz 33220A for around $500

Can you show us a picture of the broken hybrid?
I don't have a 33250A, So I have never seen inside one.

Ah, the 33220a is only 20MHz, not 30Mhz. Bummer.


Sent from my iPhone using Tapatalk

[TheSteve]

Here is a picture of the hybrid and inside - there is no shortage components and there are more on the bottom side of the PCB, on the front panel and of course the power supply.

[james_s]

What's the other side of the hybrid look like? Any chance of repairing it? Some hybrids are far more complex and specialized than others. While it may be unlikely, it's possible a printed resistor was burned out or something else simple.

[HighVoltage]

Ah, the 33220a is only 20MHz, not 30Mhz. Bummer.

Ahh, my bad, sorry.
I mixed it up with my 33522B that goes up to 30 MHz.

But the 33220A has GPIB

[daveyk]

Here is a picture of the hybrid and inside - there is no shortage components and there are more on the bottom side of the PCB, on the front panel and of course the power supply.

Than you; it's square. I was thinking of another one from a different instrument when I described it as rectangular, but that's it.


Sent from my iPhone using Tapatalk

[daveyk]

What's the other side of the hybrid look like? Any chance of repairing it? Some hybrids are far more complex and specialized than others. While it may be unlikely, it's possible a printed resistor was burned out or something else simple.

I don't recollect being able to see anything. I have nothing to lose, so I guess I could pull it and look. Thanks for the suggestion.

I am thinking of looking at the output amplifier of a Wavetek 166 and seeing if I couldn't built that on a perf board and cobble it in there. Yea, it wouldn't have the performance it should but it might work. I guess no one has ever really gotten past a fried hybrid in these things.

I did have some good news this morning. I finally fixed my Agilent E3632a power supply from the same dumpster (another thread). After replacing the bridge rectifier and both pass FETs, it still deregulated at 14volts. So thyristors showed up yesterday. I replaced them. No fracking luck. This morning something came to me in a dream so I checked it out when I got up. The wire connection from the transformer tap to the pcb was bad. That transformer tap read 6k to the other taps. The crimp connector was bad! I soldered it and now almost zero ohms between taps and the PS works perfect. Good Lord, several weeks and about $50 in parts and it was a bad wire crimp!

So that proves it; I'm dumb. I scoped and metered that thin for several weeks and couldn't figure it out but my subconscious in a dream could. I need to hire my subconscious full time.  (Sigh).

Dave


Sent from my iPhone using Tapatalk

[daveyk]

Ah, the 33220a is only 20MHz, not 30Mhz. Bummer.

Ahh, my bad, sorry.
I mixed it up with my 33522B that goes up to 30 MHz.

But the 33220A has GPIB

No problems; that's still a great idea, but I'm not seeing any on eBay yet. I'll keep looking. It would probably work fine for me and be a bit less money.

Dave


Sent from my iPhone using Tapatalk

[daveyk]

Ah, the 33220a is only 20MHz, not 30Mhz. Bummer.

Ahh, my bad, sorry.
I mixed it up with my 33522B that goes up to 30 MHz.

But the 33220A has GPIB

I guess if I have to, I could use an HP8116a. The output is not accurate compared to what you tell it to be, but the scope can monitor it and then I can have the program send commands to tweak the generator.  I think that will be a last resort, but they are dirt cheap.

P.S. - I don't know if HP-IB is compatible with GPIB, so that might not be a good idea.


Sent from my iPhone using Tapatalk

[james_s]

HP invented HP-IB, later other manufactures started using it and the name changed to GPIB, it's the same interface.

[SaabFAN]

Once it is known what the exact job of this hybrid is, I don't see why it should be impossible to recreate its function on a piece of FR-4 PCB with components available from digikey.
It will never deliver the same performance as the hybrid, but it'll at least save the entire waveform-generator from the scrapyard.

Is there a pinout available? Or can one be determined by probing around in the unit? If the rest works, the incoming signals should give a hint to what is happening on that hybrid.
Also a picture of it might be helpful to see the components
Or am I completely mistaken and there's such high precision required in the manufacture of this thing that it's impossible to copy its function with anything less than ceramic substrate and laser-trimmed resistors?

[james_s]

It may even be possible to duplicate the original performance with some care. What required a hybrid in the past may be reasonable to do on a standard PCB with modern components today, even if you have to hand select parts. Then again it may use custom silicon which makes the prospect of replicating it far more daunting.

[pigrew]

Is there a pinout available? Or can one be determined by probing around in the unit? If the rest works, the incoming signals should give a hint to what is happening on that hybrid.

The service guide has the pin-out. It is a 10x gain, >80MHz amplifier, that can drive a 100 ohm load. There is a extra pin for adjusting the DC offset. It should be somewhat doable.

[james_s]

Certainly sounds possible, not trivial but a good analog designer could come up with something that would do it I'm sure. Careful attention to layout will be required.

[daveyk]

Certainly sounds possible, not trivial but a good analog designer could come up with something that would do it I'm sure. Careful attention to layout will be required.

It's just an amplifier, but there must be an awful lot in that small package. I was thinking of using the output power amplifier of the Wavetek 166 as a guide but when I looked at it, there is a huge amount of parts and it would be a huge build. There should be room inside that case if it did go bigger.

Apparently, this is a well known failure of this excellent generator. Why Keysight won't sell the part, I dunno.  If it was a $1000 generator I could almost understand, but a $6000 instrument and the only repair option is a $3500 swap-out?

If the problem is that prevalent , someone could design an adequate board and sell it. I doubt it would be worth anyone's time or investment.

I could bring the "good" signal to the front panel out of the internal attenuators (which are before the power amp, or the majority of them are), if there was an off the shelf RF amplifier one could buy relatively cheap, but that's probably too much of a cobble.

Just a nice, simple amplifier/driver would be nice. Myself, I don't need over 30Mhz. Output amplitude variance can be compensated for.

I'm sure this is a lost cause and I will just buy a used one (with return policy) off of eBay for about $2500.00. The other generator mentioned might be an idea too.

Since they can't be repaired, even a Chinese brand in the $1000 range may be an idea and write some code for it, but they don't do GPIB, probably just network and the rest of the test equipment I use is still GPIB.

I need to do some more searching....


Sent from my iPhone using Tapatalk

[daveyk]

Certainly sounds possible, not trivial but a good analog designer could come up with something that would do it I'm sure. Careful attention to layout will be required.

https://www.ebay.com/itm/300331081153

There's a possible alternative and it even has GPIB.

Does anyone have experience with RIGOL? That may be early RIGOL.



Sent from my iPhone using Tapatalk

[TheSteve]

Do you require a floating output? Very few "off brands" offer that.

[daveyk]

Do you require a floating output? Very few "off brands" offer that.

Probably not because I'm not sure what you are referring to. I use a trigger, or Gate, input, sinewave 10mv to 10vPP. A built in accurate delay function is nice, but not a deal breaker.

Dave


Sent from my iPhone using Tapatalk

[daveyk]

Do you require a floating output? Very few "off brands" offer that.

Okay, DC offset? Not really unless when using a pulse or square wave output.


Sent from my iPhone using Tapatalk