rca: Ch= RC-1038; 66X3

I recently purchased an RCA 66X3 AM/SW reciever from eBay. It initially had the normal bad capacitors and several >20% out of tolerance resistors. Tubes tested OK with perhaps the RF amp (12SG7) a little weak but good. Tune up seemed to go OK, but the SW band was definitely poor with a lot of extraneous mixes on the dial that appeared to be from nearby FM stations. Sensitivity and volume was very low.

I have a battery operated o'scope and discovered that the LO seemed to be on the AVC circuits and on the circuit ground of the 12SQ7 resulting in LO on the plate of the RF amp. This situation only exists while in SW. The regular AM broadcast band is actually quite good and there is no LO on the RF amp or AVC.

I am wondering if anyone else has experienced this problem and what they did to correct it. I have had some emails with other less technical owners and they state that their sets are also weak on SW but I don't have any information on strange mixes in their sets.

Update: I found that bypassing the circuit ground of the 12SG7 (B-) to chassis did not affect the level of LO signal on the plate. LO signal on the pin 8 (signal in) of the 12AS7 (mixer) is 180° out of phase with the LO on pin 5 (osc). This must have something to do with the band switch, but contacts appear to be OK. The band switch shorts out a section of the oscillator coil to B- rather than switching in a different coil. B- comes to the bandswitch via an insulated braided cable (like the shield off of a piece of coax) and LO (12mHz) signal can be seen on the cable. Once the LO signal is coupled onto the AVC circuit, it is present on the grid of the 12SG7 (RF amp) which makes this one of those "chicken or egg" questions. 

To thank the Author because you find the post helpful or well done.

hi, I think we should go step by step

- is C7 ok? what happens if you connect a parallel cap (low loss - ceramic)?

- is there a cap from common to chassis?

- signal at common against chassis?

To thank the Author because you find the post helpful or well done.

Thanks for responding Konrad. C7 has been replaced along with all of the paper caps and electrolylics.  Most of the paper caps were bulged out the ends and there were pools of wax on the cabinet floor, so I thought this was a good approach. R4 (220k) and C3 (0.1uf) connect B- (common) and chassis.

After doing some back-tracking, I think that this small amount of LO on the AVC line is normal even though I'm not sure I like it and it seems to be present in AM or SW. After disconnecting C2 (RF amp coupling cap to Mixer) I found that the V2 G8 voltage was about -0.9v while AM was selected and -2.0v when SW was selected. LO signal on V2 G8 while in SW was still present at about the same level as if the coupling capacitor was connected. So the junction of R5 and R2 goes more negative while in SW while the junction of R5 and R12 does not change and is around -0.5v (no station) in both SW and AM.

Measurements of the LO at V2 G5 was about 40vpp with a bias of -5.2v while in AM, and 15vpp with a bias of -4.4v while in SW. I think that this large difference may be more of the problem than a little LO on the AVC line. 

All of these measurements were taken from B- not chassis.

To thank the Author because you find the post helpful or well done.

I still do not trust C7. What happens if you parallel a ceramic type? (or is it already?) .And/or a higher value?

Hypothesis:
Anyway either the oscillation level is too low due to component loss/ coupling, or there is kind of compensation feedback from AGC line to pin8 of V2.
What happens to the amplitude at pin5 if you shortcut pins3+4 (=D+C of 1st IF transformer)

That would indicate if the effect is caused by compensation or by "poor" circuitry/layout.

The fact that LO is counterphase between pin5 and pin8 of V2 points to a compensation effect, since G8works throug the cathode against the correct oscillation circuit.

What I do not like at all: the complete osc. and tuning return is referring to AGC while screens and plate are referring RFwise to B-. That's why I am still supecting C7 to play a role.

To thank the Author because you find the post helpful or well done.

Konrad, sorry for the delay in responding. Sometimes the hobby has to take a back seat. And the honey-dos seem to get a high priority at times.

I did as you suggested. With the primary of T1 shorted the Osc amplitudes did not change that I could measure (both AM and SW bands) on V2 pin 5. I also tried replacing C7 with a very good mica with no improvement or change.

I suspected that the Osc levels were a lot of the problem and compared results on another set (unidentified Crosley model) with a 12A8GT for the Osc and got very similar numbers for the levels of LO - about 40vpp on AM and 15-16vpp on SW). So this part seems pretty typical at least based on two sets with similar tubes. The Crosley was also quieter on SW. 

I removed the R2 connection from the AVC circuit and observed V2 pin 8. While in AM this osc level was 900mvpp with a -0.88vdc mean and in SW it was 2.4vpp with a -1.68vdc mean. The AVC voltage stayed at about -0.8vdc in either band (no station). So the AVC voltage jump (junction of R5 & R2) seems to be sourced from pin 8 of V2.

While I would like to see that LO drive be higher on the SW band, I'm not sure how to make a simple change to make that happen. So, I ran about 10 meters of wire around the top of my room and hooked it up to the antenna input. This made a big difference in the SW band obviously, so I am assuming that the built-in loop is not that great on SW. 

Conclussion: I feel somewhat like the boy who cried wolf. I think this design is doing about the best it can with the tubes that I can locate. I don't like the misc mixes I get from my location being near some FM broadcast stations but I would bet that relocating several kilometers further away from them would help a lot. With the external antenna, I do get radio Havanna pretty well as well as many other actual SW stations across the entire band. The volume is improved but still not very loud. Clear and understandable and may sound louder once in the cabinet.

To thank the Author because you find the post helpful or well done.

I would agree to your conclusion, that the design is a bit weak.
There is of course no reason to assume always an optimized product.

If so it is a good plan to use it as it is. To improve things beyond their original performance is a ambiguoos idea. It is of course challenging to make it better than the desinerwas able or permitted by alimited cost factor.
Onthe other hand it is good to respect the historical values. Everybody can DX using better radios without spoiling an original.

Enjoy it,
KoBi

To thank the Author because you find the post helpful or well done.

I did a final tune-up that helped a bit more with the sensitivity. Tuning of the SW band with respect to the oscillator tracking was very critical and it took quite a while to find the 'sweet spot' on the L10/C20 combination as well as the L14/C26 antenna tune adjustments. Each iteration of L10/C20 required L14/C26 peaking as well. At first I could not get good tracking until I added a 15pf mica across C20 as C20 was always at maximum capacitance and wanting a bit more, then after working for quite a while I was able to get the 9.5 and 11.8 MHz points to line up. 

All of this did improve the volume and it is satisfactory with a wire antenna. Perhaps one of my winter projects will be to build and install a folded dipole in the attic space for future use. Might have just enough length up there.

Thanks for your help and patience. After working with solid state for so many years, I forget the limitations of some of these old home radios. They are fun though.

To thank the Author because you find the post helpful or well done.