AC74 divider problems.

[Chris Wilson]

I have built a couple of Softrock Lite II IF frequency receivers, normally they would be used to view another receiver's IF on a bandscope, but popularly they are also used on low and medium frequencies as stand alone SDR receivers. The MF one works fine. Th LF one was intermittent, then failed with high current draw. I isolated it to a faulty or shorted divider chip. The original one was marked 78C49NM. I couldn't find one of those, so plumped for one from RS Components here in the UK, a CD74AC74M96G4, RS part number 662-6939 The oscillator worked on the bench and from an xtal frequency of 461.5 kHz gave the expected divide by 4 output of 115.364 or very close. Then suddenly it started receiving broadcast stations and the divided outputs Q and /Q were on 153.814 KHz! There they remain. Fiddling I briefly saw a return to 115.36 Khz, but it was not for long. All voltages to it's pins are correct at 5V. I am not sure if it's a faulty divider, or a bad connection or short.

Xtal case is grounded and seems stable. R16 in the oscillator changed on advice from 22.1k to 10k as they apparently tend to stop running with 21.1k there. I attach the schematic.  Wrong IC? Details of the dider's operation are at http://www.wb5rvz.org/softrock_lite_ii/03_div

The new IC details are at https://uk.rs-online.com/web/p/flip-flop-ics/6626939/

I see these frequencies on these pins:

Pin 3 : 464.444 kHz

Pin 11 : 464.444

Pin 12 : 153.814

Pin 8 : 153.814

Pin 2 : 307.625

Pin 2 : 307.625

Is this a random IC fault or does it have some mathematical correlation? My maths is pitiful!

Not sure what to do now, I have another IC but before risking pad damage thought it prudent to ask, many thanks.

Pins 14 and 2 on the mixer IC show good, 90 degree out of phase square waveforms of normal amplitude on a scope, but at 153.814 KHz....

[Andy Watson]

The Fairchild datasheet for the AC74 specifies a "minimum edge rate" of 125mV/ns - I take this to imply that the chip might misbehave if you don't meet or exceed its \$\Delta V / \Delta t\$ rate. How fast is your rising clock-edge? You might do better with the slower versions of the 74 (LS74 ? can't remember).
Also, there's an underwhelming presence of decoupling on the circuit.

[David Hess]

You might do better with the slower versions of the 74 (LS74 ? can't remember).
Also, there's an underwhelming presence of decoupling on the circuit.

That would be the 74HC74.

[edavid]

Probably you don't have enough output from the crystal oscillator to drive the divider properly.  Can you borrow a scope to look at it?

Change R16 back to 22K.

C12 is too small for this frequency... try 220pF.

You may need to change C10 and C11 as well.

Also, you could try using a 74ACT74 instead of a 74AC74, since it has a lower threshold.

[David Hess]

R16 in the oscillator changed on advice from 22.1k to 10k as they apparently tend to stop running with 21.1k there. I attach the schematic.

R16 should not affect the oscillator but it will affect the biasing of Q2 which should be held just barely in cutoff.  It might be beneficial to increase the value of C12.

Instead of a single transistor for Q2, I would have used a differential pair, logic gate, or comparator to true up the clock signal.

If you do not want to mess with the design that much, then use an oscilloscope to monitor the collector of Q2 and adjust R16 and C12 for a waveform with the fastest edges.

[Chris Wilson]

I should have said thet running on 136KHz and was not in the original design, a  1.856 MHz crystal was what the oscillator in the schematic was designed for, so i altered it from the get go to this design:

http://blog.n1bug.com/2017/02/01/2200-meter-softrock-lite-ii/

N1BUG has had no issues with his but i am running old crystals from people's "junk boxes" although NOS. I wonder if the values in the redesign look better, which is what I am using? I should have made that clear in the original post but didn't suspect the Colpitts itself. I will look at the oscillator output again with the scope and more carefully. I am loathe to change the divider IC again in case tracks or pads lift. R16 change and crystal change is less fraught :)Thank you all for the help, as always

[Chris Wilson]

I just completed two scope measurements t R17 to pin 3 of the divider IC. One of the faulty LF Softrock, and one on my MF one that works fine. the resultant wave forms are quite different. But I am not clever enough to say if they hold the clue, so will post them here. Thanks again.

Faulty LF oscillator or divider first

[edavid]

Is your scope AC coupled?

You need to know the actual high and low voltage points of the waveform.

[David Hess]

What we are seeing there is that at low frequencies, the value of C12 and possibly the values of C10 and C11 needs to be higher.

N1BUG has had no issues with his but i am running old crystals from people's "junk boxes" although NOS.

I do not think the problem is with the crystals so much as the oscillator design being more sensitive to component variations than it should be.  For instance it lacks temperature compensation of the operating points and it was not designed to operate at such a low frequency.

I am loathe to change the divider IC again in case tracks or pads lift.

I doubt there is anything wrong with the divider beyond the possibility that the input waveform does not meet the maximum transition time requirements for the clock input.

[Chris Wilson]

Thanks all for the replies. I managed to change the divider again, no real change. Playing with R16 helped a bit, but basically the output is unstable from the divider. Just putting a finger very near the divider plastic body will get it going, but often the square wave forms go out of 90 degree phase shift and dither about. I am wondering if such a low frequency oscillator could be bread boarded and optimized somehow on there rather than pulling stuff off the board with a real risk of damage? I will try 220Pf forC12 though, thanks edavid. can anyone suggest values for C10 and C11 that should work, is there a way of calculating them?  I am not sure if I have any spare 2n3904 and 3906 devices, I may have to buy some unless they are very none critical? I'll also put back the 22k for R16 if you think I am changing it needlessly. Thanks for all the advice, this is the waveform on R17 to pin 3 I now see.

Andy, if I build on a new PCB should I either try and get exactly the same divider as the kits came with, or go for a slower one? I don't want to change too much willy nilly as I am easily confused!

Have a good weekend. I was hoping this simple RX would be a walk in the park, I am managing to get my home brew TX set up into the US as I type, I am embarrassed that I am struggling to get a simple SDR RX running right!

[edavid]

That waveform looks fine.  What are the waveforms at pin 6 and pin 8?

[Chris Wilson]

That waveform looks fine.  What are the waveforms at pin 6 and pin 8?

I didn't capture those, they were fickle, when they stabilized the frequency was correct and they were sharp edged square waves, 90 degrees out of phase, but the rest of the time they were either none existent at all, or jittering in phase. Looked like it might have been oscillating. I basically didn't have enough hands to hold the probes, work the keyboard and all that stuff I can't try again as I have pulled C12. I have just realized I missed off a link to the mods to the basic 455Khz circuit I worked from, to run at 136KHz, which show the new values of C10, C11 and C12, but I needed to check C12 as I got it hot fitting it, as it is tiny and awkward to solder. It would have been the value in the mods sheet though, but I will replace it, but await comments on these values. Thanks again

The mods I am working to are at:  http://blog.n1bug.com/2017/02/01/2200-meter-softrock-lite-ii/

[Chris Wilson]

Today I decided to try two things, build a little test rig with a spare xtal of the same 461.5kHz as the one in the RX with the same C10 at 3900pF and C11 at 2700pF on a bit of pcb and excite it at 1v P to P sine wave from my sig gen around the xtal frequency and look at the output on the scope. That failed abysmally, no change in display, thinking about it maybe it's due to the big capacitances involved?

Anyway, back to the RX. C12 at 680Pf, C10 at 3900pF and C11 at 2700pF. Waveforms taken with scope at DC coupling, but V scaling changes. I believe (dangerous with my lack of knowledge..) that C10 to base of Q1 is good. Base of Q2 looks ok. But with R16 at the recommended 10K I see half a sine wave at the emitter of Q2. With the original 20k (I don't  have the oddball 20.1k around) the waveform is still part complete and of a much lower amplitude.

Is this a bias issue with Q2 or something else please?

I will have to split the files over 2 posts,, due to number and size, sorry.

I hope the file names are self explanatory! Cheers.

[Chris Wilson]

Rest of scope screen shots