diy IF transformers: first attempt.

[ppppenguin]

Now we know the testgear, how are you injecting and sampling the signals? Bearing in mind this can have a big effect on the result.

[Amie]

Now we know the testgear, how are you injecting and sampling the signals? Bearing in mind this can have a big effect on the result.

Hi Jeffrey.

OK, so, function generator set to a max frequency of 490kHz and a sweep width of 50kHz, output checked at this by the counter. Repetition  of sweep for these settings comfortable for the scope horizontal tb set to 0.1mS /div. Output amplitude over this small range is practically constant, and set at 1v. Scope set for vertical amplitude of 10mV/div.

Both "windings" of the experimental IFT are identical, but input fed through 40k source resistance and output to the scope by way of  ×10 probe, (resistive loading at 1M not a problem but input C of ≈20pF knocks it off tune badly, obviously any in circuit strays and loading will have to be allowed for as normal). I would have liked to have used 49pf or 68pf fixed C and 100pF trimmer but although I had some new ones, 9 of the 10 were faulty

Thanks,
Amie.

[ppppenguin]

In many applications at anything more than pretty low frequencies the capacitance of a scope probe usually matters more than the resistance. The 40k source should keep the loading on the circuit pretty low.

If your scope has enough sensitivity you might like to try a low resistance passive probe. This is often a 4k7 resistor soldered directly to the point under test, with the signal fed back to the scope via 50R co-ax and terminated in 50R at the scope. This method works well to beyond 1GHz and is excellent for probing high speed logic. Of course you're getting 100:1 attenuation. If you can stand even more attenuation than that you can use higher value resistors. There's no point in going that high resistance for very high frequencies but it might just give a good result in your application.

[Kalee20]

Though even if you accept x1,000 attenuation, and use a 47kΩ resistor and 50Ω coax and termination, you still load the circuit under test with that 47kΩ, which will affect the Q and characteristics. IFT is meant for driving into a open-circuit grid, after all.

It's a good technique, just not applicable here.

Amie's x10 probe with 1MΩ loading (or 10MΩ ? as most x10 probes are) is more suitable in this scenario. Yes, there's capacitance, but that can be allowed for by reduction in trimmer.

Or, (better), a JFET source-follower powered by local 9V battery, the JFET gate lead tacked directly onto the Circuit Under Test. And then you really can have 4.7k and 50Ω to the scope!

[Amie]

Hi

Thanks Jeffrey, and Peter.

Just to clarify, scope IP is 1M and 20pF. The 1M not really a problem the capacitance is difficult to dial out at 455kHz with my  trimmer and fixed caps, no matter I think I have a good enough design with two identical devices to go from FC output through 1 IF amplifier pentode and drive an infinite impedance detector.... (AGC to yet be considered! )



Amie.

[pamphonica]

What a nice experiment, and clear results. I might use it at the radio club one day.
Just a thought, but would this method make the transformers a bit microphonic? Tapping on the case of the radio would vibrate them and that gain variation might feed through to the audio. Maybe they could be fixed to a thin fibreglass support with epoxy or nail varnish once the spacing is established. Any thoughts anyone?
-Jeremy.

[Amie]

Hello Jeremy.

As you've noted the coupling is altered by bending the coils toward and away from each other, the epcos ones I used (from RS some time ago) are quite sturdy though but I had thought of trying an old technique of putting a small piece of soft foam between them after optimising the adjustment then fixing the foam by soaking with hot beeswax....

Also....

Cut and paste from a PM where I was asked to clarify what load the coil was working into....

IF transformer testing and oscilloscope input loading.....
I think I must have made a pigs ear of explaining I made the transformers with the 820microhenry chokes with a fixed 100pf//50pf trimmers. The scope input is 1M//25pf (I actually said 20pF ), but that made the transformers unable to tune "up to" 456kHz as I'd wished due to the input capacitance of the 'scope, therefore I used the probe not on times 1 but times 10, so not only giving a load capacitance of ≈2½pF enabling the transformer to tune (?) But also ≈10M load impedance.

Amie

[ppppenguin]

Good explanation but unfortuantely the probe tip capacitance doesn't directly scale with attenuation. This is mainly due to the capacitance of the probe cable. A x10 probe usually presents 8 to 10pF at the tip. x100 probes (which are quite rare and often expensive) can get down to 2pF or so. Active FET probes can go even lower but they are electrically fragile and horribly expensive.

The capacitance at the tip of a x1 probe is often about 100pF, again mainly due to the cable. For almost all purposes x1 (and switchable x1/x10) probes should be banned. Nasty things that will mislead those who don't understand.

[Amie]

Good explanation but unfortuantely the probe tip capacitance doesn't directly scale with attenuation.....

That would explain the tuning problem even better if the case.

....The capacitance at the tip of a x1 probe is often about 100pF, again mainly due to the cable. For almost all purposes x1 (and switchable x1/x10) probes should be banned. Nasty things that will mislead those who don't understand..

I think I may have possibly just proven that point Jeffrey!

Thanks,
Amie.

[Kalee20]

For almost all purposes x1 (and switchable x1/x10) probes should be banned. Nasty things that will mislead those who don't understand.

x10 probes can be misunderstood too, I was using a x10 probe with a scope in the presence of the Quality Manager at a company. First thing I did of course was, clip it to the 'cal' square-wave output on the scope, the displayed waveform showed the probe compensation was wrong, so I adjusted the probe trimmer, which was covered by a label, for a nice flat-topped square-wave.

The QA manager watched me in horror. "You've just invalidated the calibration of that probe!" he said.

Words followed... And the QA manager didn't last much longer...