28 September 2007

Closed for holidays

About to take a short vacation, with lots of open projects:
  • FT-817 hands free adapter
  • Finishing 4m TX #1
  • Rebuilding 4m TX #2 (it has a short Vcc-GND on the ICS512 adapter)
  • Put into an enclosure the NiMH constant-current charger
  • Adapt two relays for 7A13 Tek plugin
  • Put into an enclosure the Cisco 7960 headset adapter
That's it for the circuits laying on the workbench. Pending circuits list is much longer...

25 September 2007

Controlling canned oscillators

For a number of projects it might be interesting to control the frequency of a canned oscillator. Some of them have a control pin, others an enable pin, others... no option.

There are two possibilities, one of which should always work:
  1. controlling the voltage supply
  2. controlling the load capacitance
The latter works only if the module does not include an output buffer. Most recent canned oscillators should be insensitive to load C variations.

The voltage supply control apparently works on all canned oscillators. You'll need to map your oscillator behavior vs. voltage, but a few Hz shift can be achieved, say 200Hz at 30 MHz, that is 0.06%.


17 September 2007

Tektronix 7A13: broken relays

Relays sit on sockets, so it's very easy to test them outside the original location. Pinout is printed on one side and, before you worry, they're reversible. Surprise: they have 8 pins per side, but row spacing in just 1 pin, so they don't fit on the breadboard as an IC would. So I built an adapter...

If you notice my adapter has 4 pins on one side and only 3 on the opposite: it's because I ran out of pins. But these Tek beauties are reversible, so 5 pins would be enough.

Result of testing 2 relays: out of total 4 switches, they all show continuity in the rest position and only one in the switched position. All others do "click" but don't close the contact. #@£!!

These DPDT relays are Tek 148-0034-00, 15V 600 ohm. No chances to get a direct and new replacement.

But the 7A13 plugin has two other relays with the same part#, used to reduce the bandwidth to 5 MHz. The default position (relays in "rest") is "full BW", that is what I am more likely to use (RF vs. LF). I did not check these K480 and K490 for proper operation, but a self-transplant restored 2x and 5x V/div ranges.

If no other relay is available, use a short jumper to close the contact of interest and live with it.

14 September 2007

Tektronix 7A13, debugging

I have a Tek 7A13 amplifier plug-in that shows a problem. The 1V:100mV:10mV/div ranges work fine, but others don't.

With both inputs at GND, trace centered on the screen, changing to a 2x or 5x voltage range brings the trace down a couple of divisions. If a signal is injected in 2x or 5x ranges, nothing happens. A signal can be injected this way:
  • bring to (+) 0.000V the internal voltage generator
  • set both inputs to GND and POSITION the trace on mid screen
  • set IN+ to Vc: you're now comparing the voltage generator to GND
  • slowly increase the voltage
If the range setting is working, the trace will move up and down together with the voltage. Return the generator to 0.000 and change range: if the trace jumps, there's a problem.

And here the original paper documentation becomes useful. Since only three ranges work on my unit, I started my search from board 1, where the selector is located.
Amongst all poles, there's one that drives two relays, namely K47 and K48, in all ranges except those that work. They control a resistive network on board 3. Ouch!, BTW, this is also described in the "Circuit Description" section of the manual!!!

So there is the possible culprit: a relay, a contact on the rotary switch or a connecting cable between boards.

I will inspect it soon...

07 September 2007

4m TX - foto

Ecco come si presentava il TX il 6 settembre 2007:

Da sinistra a destra: filtro C-L-C, BNC e rele', 74AC08, moltiplicatore x5 e, in alto, il regolatore a 5V.

Il rele' nella posizione di riposo (non alimentato) gira il segnale del BNC verso l'RX. Questo connettore di uscita verra' aggiunto quando il TX sara' inscatolato. Peraltro tra il rele' e il BNC vorrei inserire un filtro passa-banda o passa-basso. A circuito acceso il rele' manda in antenna il segnale del trasmettitore.

Il filo nero che si infila sotto il quarzo e' un rudimentale tasto CW. Il quarzo lo tiene a massa ("key up").

Il rele' e' collegato direttamente all'alimentazione a 12V, dato che usando un RX separato l'oscillatore deve essere spento durante la ricezione. Ecco perche' a riposo il contatto ruota l'antenna sul connettore che ancora non c'e'.

Comb Generator - measured

I built the circuit in about one hour. With an accurate selection of NOR gates usage, the signal can be routed through the 74HC02 simply bending pins inwards:

This build is a preliminary version to verify my assumptions (the theory). I used a 14.000 MHz XTAL and I didn't care to keep leads short or add bypass capacitors: the more noise the better.

The 100 MHz frequency counter could measure the 14 MHz signal at the oscillator but went crazy when picking up the output, showing 3.8 MHz or so. Perhaps the glitches were too fast?

On the spectrum analyzer the output looked like this:

45-150 MHz, marker 70 MHz

150-450 MHz, marker 210 MHz

Comb teeth are 14 MHz apart (XTAL frequency, not XTAL/2 as stated in a previous post). Measured harmonic levels are as follows:

MHz dBm
42 -14
56 -14
70 -13,4
84 -14,4
98 -12
112 -15
126 -13,3
140 -17
154 -16,5
168 -18
182 -20
196 -21
210 -23,3
224 -28
238 -28
252 -34
266 -35
280 -43

Note: 42 MHz is estimated.

Up to 150 MHz levels are quite constant. The further quick decrease is probably amplified by the attenuation of RG58 cable used for measurement.

I cannot explain the notch at 294 MHz (lambda ~= 1 meter), maybe some series resonance in the measurement setup? Or an expected zero on the spike spectrum? Higher harmonics are between -35 and 40 dBm up to 450 MHz.

Conclusion. Although the comb is obviously not leveled it is useful for comparative measurements on VHF/UHF filters:
  1. take note of levels without filter
  2. insert the filter between the comb generator and the spectrum analyzer
  3. tune the filter and measure harmonic levels and compare with step 1
Add-on. A small trimcap on the output of the delay gate may vary the output duty-cycle, thus modify the theoretical spectrum roll-off in case measurements fall in the ~300 MHz range.

06 September 2007

Comb Generator - simulated

After checking few times the paperwork with logic states to confirm the expected behavior, I tried simulating the comb generator circuit.

I used NOR gates instead of NAND because I will build it with a 74HC02 and not a 74HC00, but the effect is the same.

I found a Java-based logic simulator called logsim. It does not produce output vs. time plots, so a D flip-flop is added to evaluate the effect of propagation delay (as described in logsim manual).

Here is the diagram. Don't forget to RTFM.

Results so far: in theory, it works. Simulated, it work. Will it work?

05 September 2007

Comb generator, preliminary

This is a preliminary circuit diagram of a comb generator using a 74HC00/74AC00:

The first gate is biased as a XTAL oscillator. I picked 10 MHz but any HF junkbox rock will probably do. Pretty standard stuff.

The second gate is a simple buffer for the oscillator output. Makes sense.

The third gate delays the signal of few nanoseconds, enough to produce -hopefully- a glitch in the 4th gate output. Will it work?

The R-R-R network is a simple ~10dB ~50ohm attenuator pad to match the impedance to 50 ohm loads (D.U.T. - Devices Under Test - mainly HF/VHF filters in my case).

According to my paperwork the comb peaks should be separated of half the XTAL frequency.

I've always wanted to put to a good use TTL gates propagation delay! Let's see...

Filtro passa-banda per i 70 MHz difettoso

La realizzazione presenta una attenuazione di 80dB alla frequenza fondamentale. O c'e' una saldatura fredda o ho sbagliato qualcosa.

Primo passo di debug: rimuovere uno dei due condensatori da 100pF usati per arrivare a 212pF. Forse il compensatore usato (fonte: cassettino dei recuperi) e' di capacita' troppo elevata.

Se la situazione non migliora abbozzo al volo un filtro passa-banda LC riciclando i due C da 100pF e l'induttore di cui gia' dispongo:

Se proprio non ne esco, ripristino l'uscita dell'antenna a valle del primo filtro originale e controllo che la potenza sia quella di prima (21dBm).

C'era una saldatura fredda! Diamine! In realta' si era staccato il collegamento tra l'uscita della porta NAND e il primo LPF. Ovviamente avevo gia' fatto tutto il debug indicato sopra.

Il "finale" non si e' bruciato, e infatti all'uscita del primo LPF ci sono ancora 21 dBm.

Alla fine ho lasciato solo il filtro BPF ad un polo indicato nella figura ed ho ottenuto i seguenti livelli:

[MHz] [dBm] [dBc] [dBm] [dBc] [dBm] [dBc] [dBm] [dBc]
70 14 0 14 0 21,5 0 13 0
140 -35 -49 -38 -52 -19 -40,5 -45 -58
210 -40 -54 -40 -54 -19 -40,5 -46 -59
280 -52 -66 -49 -63 -26 -47,5 -69 -82
350 -47 -61 -35 -49 -24 -45,5

470 -50 -64 -56 -70 -28 -49,5

La potenza e' scesa, di moltissimo. Il filtro al momento non e' sintonizzabile... urge costruire un "comb generator" per misurarne l'effettiva risposta in frequenza.

04 September 2007

Cosa manca in un'auto

Hai acquistato un'auto nuova o usata ad estensione del parco macchine, non in sostituzione di un veicolo: cosa le manca per renderla confortevole?
  • giubbottino catarinfrangente (obbligatorio per legge)
  • ombrello
  • stracci per i vetri
  • tappetini
  • parasole da parcheggio
  • carta e penna
  • cartine stradali della citta'/zona (se non si dispone del navigatore satellitare)
  • auricolare del cellulare
  • (supporto per) telepass
Se si e' radioamatore:
  • base magnetica
  • antenna
  • cavo di alimentazione RTX dall'accendisigari (controllare la polarita'!)
Se si hanno bambini:
  • seggiolino/rialzo (obbligatorio per legge)
  • giochi
  • tendine parasole
E probabilmente altro ancora...