Hewlett Packard 8601A sweep generator.

[agardiner]

Thanks Paula, I really appreciate the advise and help.

So regarding the modulation; I was looking at the RF output directly on my scope. I have however now concluded that the lack of apparent modulation on the trace is probably due to the current lack of ALC, as I note from the SM that AM modulation goes through this. So I am hoping this will be resolved when I receive the new output hybrid.

Yes, my new scope is a DSO, and the X-Y mode is pretty poor compared with old CRT ones I have used, and besides the sweep output is a ramp. So using your advice, I would expect to see the IF filter bandwidth on the channel connected to the FM detector? Would I still see a marker on the trace from a 2nd oscillator? Sorry for the ignorance, I just haven't used this kind of setup before, and have been struggling all day.

Thanks again.

[frsimen]

Hello Adrian,

The FM detector output will show the S-curve. The IF filter shape will be seen at the input to the detector. Cossor service sheets suggested using a diode probe for displaying that.

The marker should work as you would expect, you will need to play around with the marker signal's level for best results.

If you can hear the modulation, but can't see the envelope on your oscilloscope, you've either got FM rather than AM or you haven't set your oscilloscope correctly. There may be an "Acquire" menu on the DSO. Try setting that to peak detect if that is one of the options.

Paula

[Radio Wrangler]

If you want to cheat, you can always look for a modulation output or sweep output on your signal generator which you can feed to 'ext trig' input on your scope.

Sweep output works best with the scope in X-Y mode but you will see the retrace, hopefully dimly due to higher velocity on the screen.

David

[agardiner]

Thanks again Paula. All very helpful. I will wait for the new hybrid module and then carry out some more tests.

[agardiner]

Thanks David. I think at least some of the issue is my choice of DSO scope. The X-Y mode is not great to say the least. That said, I am going to wait until my new hybrid part arrives so at least I can confirm that the sweeper is in full working order before I try again.

[agardiner]

Hi Paula,

So, part has not arrived yet, but I carried out some more tests using your suggestions this afternoon, but on a known working radio so as to eliminate that part.

X-Y is totally useless as discussed. Using 2 traces, one from the sweep output and the other from the detector of the radio, and triggering the scope on the sweep; as expected a very good S curve. So that part is great.

The bit I am still unsure of is the IF bandwidth. So far, no amount of testing has produced anything like an expected curve, and impossible to see any marker. So far, the best result I have found is to switch on the Math function and set to divide the 2 signals. That produces a nice peak which seems to me to be the centre of the IF trace. If I centre the sweep waveform, then the math trace peak was showing centered on the scope, and adjusting an IF transformer did move it.

Would this be a good way of working, or am I still missing something? Certainly X-Y is not going to be an option.

[agardiner]

OK, so I think I have sussed it. Played with some settings and slowed the sweep rate right down. I now have the expected shape of waveform, and could certainly align it, but, currently there is far too much noise. I have started a separate post about this: https://www.vintage-radio.net/forum/...d.php?t=200592

Forget the math part, I have now realised this was not the way to go. My issue is that the noise flow caused by the test equipment, is making it really difficult to dial in the waveform. I would certainly appreciate it if you have the time to check the other thread and pass comment.

Thanks as always.

[Radio Wrangler]

Swept frequency response testing into any sort of broadband detector inevitably suffers limited dynamic range. The broadband detector lets in a lot of bandwidth of whatever noise is present and it lifts the lower extremes of the plot on the screen.

The way out of this is to use a network analyser. In one of these, the signal level is detected by a superhet receiver tracking your sweep generator. This receiver has a very low IF (audio sort of region) and has no filter to prevent image 'reception'. The low IF has a much narrower bandwidth than the broadband approach whose bandwidth needs to exceed the sweep width (this often far exceeds the sweep width) So you get a much lower noise floor, and you avoid diode detector threshold effects.

Usually a network analyser will have two such receivers for comparison purposes. A more elaborate network analyser will have a third receiver using one to monitor the RF level from the sweeper and two to perform simultaneous measurement of transmision and reflection characteristics of whatever you're testing

The 8601 was part of an early 110MHz network analyser. The second box was a group of receivers with amplitude and phase comparisons. The third box was an X/Y display using a storage CRT.

I can't remember the instrument number of the receiver box, 84xx probably. Covers something like 100kHz to 110 MHZ, and is what those oscillator outputs on the 8901 are intended to drive.

You've now discovered the distortion and cramping of a plotted frequency response, when it is swept too fast. At first, the response peak shifts a little to the right (sweeping left to right, no matter whether the freq sweeps upwards or downwards in frequency) Then with fastr sweeps, the shift gets worse and the peak gets less, and it gets fatter, eventualy you have a partial response and an unrepresentative mess.

The cure, as you found, is to sweep more slowly. If you can narrow your network analyser's IF filters, your sweeps may be forced to be very slow, hence the need for a storage display.

If you're developing or aligning filters, A plot of reflection or of phase can provide clues showing where poles of the filter lie within a seemingly, boring featureless part of the passband.

The sweep speed limitation effect occurs with any form of swept mechanism whether yoou use wobbulators, sweepers or whatever. The sweep has to be slow enough to lrt the device under test's response to settle at a frequency before the sweep has moved on very far.

The sweep speed is limited by sometimes the DUT, sometimes the analyser bandwidth.... whichever is narrowest

David

[agardiner]

Thanks David for your comprehensive reply. Very useful information which makes sense.

I have also replied to your post in my other thread, with a follow up question.

Best regards,

[Radio Wrangler]

Got it.

David

[agardiner]

OK, progress. Managed to align an FM stage completely this morning. The trace was far from perfect with too much noise, (as per other post regarding DSO for vintage radio work). I have ordered a low pass filter for the scope which should clean up the trace nicely, fingers crossed.

Thanks again everyone. I will report further once the new hybrid module arrives.

[agardiner]

OK, just a little update in case anyone is still following this thread.

Today, my brand new old stock HP hybrid output module arrived. After fitting it and carrying our a full alignment of the instrument, the unit is now working perfectly. Once again a calibrated output, a clean sine wave across all frequencies (it was quite distorted before) and the sweep output working properly. Marvellous.

[agardiner]

Anyone got a scrap one of these? I have posted a quick wanted ad for a knob to finish it off.

[agardiner]

I have one further question if anyone can help please. Obviously I don't want the replacement hybrid module to bite the dust. Should I be terminating the output with a 50R load? When using on a vintage radio, the impedance is obviously going to be high. I presume this would cause unwanted reflection back into the signal generator, or is this not relevant given the low power levels involved?

[Radio Wrangler]

Have a play with your scope's trigger settings. try triggering from each channel.

Davd

[agardiner]

Hi David.

Sorry you have lost me. Can you explain? Following the repair and alignment of the unit, it is working perfectly. Just want to know if I should be terminating the output with a 50R load when using the unit. I know without it the output will be much higher than the indicated power, but what about standing waves?

Thanks.

[agardiner]

So, I have been thinking about the terminator issue. It would seem a good idea anyway, so that output level of the signal generator is correct as per the calibrated meter.

I am planning to make a simple pass through terminator, complete with DC blocking cap, and house this in a BNC male to female case I have (pic 1). My schematic would be as indicated (pic 2). I could then plug this into the signal generator output and feed the radio from the output of the device. Good idea? Or am I over thinking this?

Thanks.

[Radio Wrangler]

If you have two modulated signal generators running into two channels of a scope producing two traces and the signal generators are free running and independent, you can only trigger on one. The other will have a sliding trace, perhaps sliding past so fast that you don't see the modulation as such. Which trace shows the modulation stationary, is the one you are triggering form and have set trigger levels etc appropriately.

This is my best guess for why one sig gen seems to be unmodulated.

8601s are probably OK without a load. Most damage is from people testing transceivers and accidentally transmitting into their output.

David