Restoration of a Grundig 3033.

[YC-156]

(In this article I decided not to inline the photos. Instead the interested reader is encouraged to click the named links below. This will pop open a new browser window with the requested photo shown in 800x600 resolution. Just close the new window when done with a given photo.)

While my recent restauration of a 1956 Grundig 3033/W3D might not be considered particularly challenging, nor is the radio rare by any stretch of imagination, I decided to share it anyway. Being only the third domestic radio I have restored since restarting my valve hobby activities, I believe it is the best I have to share at the moment.

The story behind this radio is simply that I spottet it on the German ePay site, and I instantly fell in love with the exterior. After doing a bit of research on the technical details, it became evident that this type of radio probably would be quite adequate for listening to today's FM transmissions on VHF. (Hey, it is an excuse! ) The description given was basically "looks nice on the outside, weak reception, distorted audio", which to my ears sounded more like "everything crucial works, needs caring hand and restoration". The exterior was in reasonably good shape, so the rest is history as they say.

A week later the radio arrived and I instantly gave it a good looking over. Unsurprisingly it was filled with 49 years worth of dust, grit and grime. (We will get to the exterior in a moment.)

Overview of the inside
Scene from a bad horror movie?

The EL84 isn't gassy. The white shade is just the dust on the outside, which has burned to ash around the hot middle of the valve envelope.

IF cans ... and more muck

Notice the adhesive strip, which, though having since peeled away, had been glueing the EF89 to the chassis. The remains are still visible on the valve envelope. I wonder if this shows that the tube is original as fitted at the factory in 1956, or if these seals were added at a later date as part of a repair- or service job?

Below deck there was scores of those dreaded Grundig paper caps, and...

View from below
Closeup of AF output stage, seen from below

...both the EL84 cathode resistor and the 1.3K/3W voltage dropper/filter resistor in the power supply had seen better days.

More muck on the chassis, some part way through the cleanup process.

Some of you...
...must have seen tons of views like these.

By the way: As I later found out two of the three dial lightbulbs still worked, though they were rather dim due to evaporated metal deposits on the inside of the bulbs.

To try to be brief I ended up doing more or less the following work on the radio before I dared consider the restoration fairly well completed. This is pretty much the order in chich I completed each task:

*) Overall disassembled and cleaned everything and anything, and lubricated all the mechanics, including the two tone controls, rotating ferrite rod antenna, three separate dial cords and piano key switchboard.

*) Cleaned chassis topside and light reflector plate behind glass dial with watchmaker's putty, after having first removed the bulk of the muck using a soft brush and a vacuum cleaner.

*) Repaired scrates on paintwork on the rear side of the glass dial, cleaned same.

*) Cleaned all pins on the tubes with a glass brush, cleaned all tube socket contacts with a tiny, tiny dental brush and contact cleaner.

*) Tested all tubes for shorts and transconductance, all passed. The EABC80 was replaced with a specimen from the junkbox, which has perfectly matching conduction curves for the two FM demodulator diodes. The original tube, though working, had an average curve difference of a whopping 13% !

*) Cleaned all sliding contact surfaces in the variable capacitors using contact cleaner on cotton swabs.

*) Replaced all electrolytic and paper capacitors, more than 20 in total. Two new PSU smooting caps were attached to the chassis below deck using double sided foam tape intended for this purpose. The original electrolytic cap is still in place, though disconnected, for decorative purposes.

*) Replaced the metal oxide high voltage rectifier with a newly constructed one using 4x 1N4007, 4x 10nF and 2x 600ohm/7W resistors in parallel to equalize the voltage drop.

*) Replaced missing 35pF ceramic capacitor in the VHF tuner input circuitry. Seems to have been previously broken by ham fisted repair person.

*) Replaced three burned out resistors. Apart from the two mentioned previously, there was another dead one in the anode of the EF89.

*) Replaced mains fuse with one of correct size.

*) Preheated mains transfomer using DC current for about 24 hours to drive out moisture (if any) in order to prevent high voltage arching.

*) Carefully fired up the set on my 'AC gizmo', which is a combination of a 200VA isolation tranny, a 200VA variac and a bunch of fuse holders and DPDT 10A mains rated switches. Inserting the valves again one by one tested all DC voltages, currents and operating points. OK.

*) Tested mains transformer for internal temperature rise after one hour of continuous operation with all valves in place. Only around 25 degrees C! OK.

*) Repaired broken loudness contact *inside* volume potentiometer! This was a fun (and unexpected) problem to locate. The loudness contact connects through an isolating plate to the resistive track through a tiny brass rivet. The head of the rivet on the conductive track side had broken off and thus the contact could go intermittent in the presence of strong bass. That issue took a bit of headscratching to locate.

In the end I had to remove the potentiometer from the radio and drill out the 4 rivets holding it together. After finally identifying the exact problem (tiny brass pieces falling out is never a good sign) I had to find a M1.4 machine screw in brass and a matching nut, and then carefully file them down to size in order for them to clear their surroundings!

I'm more than a little annoyed with myself for not photographing the interior of the volume control when I had the chance, for the innards are worth seeing IMHO. If you think these Grundig potentiometers are made like today's cheap hobby offerings ... then you would be mistaken. The contact to the resistive surface is made through a small block of milled graphite to a spring assembly, made from phosphorbronze or a similar alloy. The graphite block is shaped, so that it balances on a 'finger' from the spring assembly, making two independent contact points to the resistive surface in the process.

The contact from the wiper spring, which rides isolated on the (grounded) metal shaft, and to the exterior world, is made through two independent contact points, which rides on a circular, silver plated brass 'track'.

Apart from a bit of wear on the contact points just mentioned, there were no other evident signs of damage, so after a bit of cleaning the potentiometer went back together with four M2 machine screws replacing the original rivets. I was fairly surprised to see that there was absolutely no sign of wear on the resistive track. Not even the faintest sign of a 'racetrack' to be seen after 49 years of service. Amazing.

*) Corrected FM audio response by adjusting de-emphasis time constant to today's standard of 50uS. I originally had it 'corrected' to 75uS untill I learned that was the US standard. Oops.

*) Replaced all three dial lightbulbs, including the one inside the Ferrite Antenna direction indicator assembly.

*) Made additional light baffle for the magic eye tube, EM85. The magic eye still has some life left in it, but the bright light from the new dial lightbulbs swamped the feeble green glow. A bit of folded dark paper and some tape did the trick. Easy to remove too if required.

*) Reconnected electrostatic tweeter after testing. Was disconnected for some reason. Seems to work well enough, though it is a bit weak.

*) Tuned 10.7MHz FM IF strip and VHF tuner using signal generator. I didn't bother with the 468KHz IF transformers, nor the SW, MW or LW bands, as I don't intend to listen much to them using this radio. All bands appears to work well enough though.

*) Adjusted FM ratio detector by aiming for maximum AM suppression on a weak AM modulated signal from my signal generator. This did wonders for the audio quality on FM.

A few photos of the radio after the restoration, inside and out, as it basks in the afternoon sun.

Sunny side up
Squeaky clean.
Well, OK, nearly so.

A new layer of fine dust has started to form on the chassis, as the set awaited completion of the restoration over summer.

In the last photo it is barely possible to see one of the new M2 bolts now holding the volume potentiometer together.

The final words are that today this radio is my 'daily driver' for FM reception. There is still a hint of excessive hissing on the high audio frequencies, so I might have to try the IF detuning trick mentioned in a recent thread on this site. Wonder if it applies to 10.7MHz IF strips too...?

Hope you all enjoyed as much reading about this little glowbug adventure as I did writing about it.

Questions or comments, please?

Frank N.

[oldeurope]

Hi, looks good!.

Yes, the adhesive strip is made at the factory. So your valves are originals.
Correct agliement of the FM IF is not possible without sweeping and
beleve me it is not only having the instruments to make a good agliement!
To agly the ratio you need a "ratio voltmeter". In the pic you can see
mine. The ratio is aglined well when you have the same reading left
and right of the center frequency (0V at center frequency).
The ratio voltmeter has a build in divider to make it possible to agly
asymetrical ratios.
The 50us can not easiely calculated because the unknown stray
capacitance is added. Be sure Grundig knows what he was doing.

Kind regards
Darius

[YC-156]

Hi Darius,

Originally Posted by Darius

Correct agliement of the FM IF is not possible without sweeping and beleve me it is not only having the instruments to make a good agliement!

Well, I did have a suspicion that the rather simplified alignment instructions I have from Grundig was somewhat lacking. I have to admit too that it was only after some hesitation that I tried aligning the FM IF. But it was one of the remaining options that was available for me in order to clean up the audio response. At least I don't believe my ministrations made it any worse than it already was.

Could you provide some additional suggestions or references for doing the proper alignment, please? Also what level of frequency deviation would be required from the sweep generator during the procedure? My signal generator can provide up to +/- 100KHz of deviation in FM mode at 10.7MHz when using an external source of modulation, and it wouldn't be too much trouble to wire up a little triangle generator in combination with my scope. But in the back of my mind I have an idea that I need even more than 200KHz of sweep to do this properly...

Originally Posted by Darius

To agly the ratio you need a "ratio voltmeter". In the pic you can see mine. The ratio is aglined well when you have the same reading left and right of the center frequency (0V at center frequency). The ratio voltmeter has a build in divider to make it possible to agly
asymetrical ratios.

I did try that method, using a pair of 43K/1% resistors as the voltage divider and combining them with a DVM. Results are about the same as when tuning for minimum AM modulation, and the latter method is easier (quicker) to set up.

Originally Posted by Darius

The 50us can not easiely calculated because the unknown stray capacitance is added. Be sure Grundig knows what he was doing.

That was something else, which had me wondering. I felt the added capacitance did help, but of course, if the IF response is totally bonkers, then all bets are off anyway. Fortunately the change is trivial to back out again if required.

Maybe I have just found an excuse for buying a sweep generator...?

Thanks for your comments so far Darius.

Frank N.

[oldeurope]

Hi Frank,
find the book fm receiving principals Cook/Liff 1968 Prentice Hall
Library of Congress Catalog Card Number: 68-20857

Kind regards
Darius

[YC-156]

Thanks, Darius, I'd better try to find a copy and do my homework.

Yesterday I did a bit of grunt work and manually swept the 10.7MHz IF response. This is what I got:

10.7MHz IF response Ouch.

As the maximum frequency deviation today is +/-75KHz I think it is fairly safe to say that this response will cause severe distortion, especially for larger volumes. Due to the pre-emphasis that means high frequencies in particular.

Additionally, since this set has a fairly typical valve lineup (ECC85, ECH81, EF89, EABC80, EL84, EM85), it doesn't have a limiter stage, and I have also now realized that the AGC is effectively disabled during FM reception! The Y axis on the plot is actually the rectified voltage from across the ratio detector, and it is certainly not the largest voltage I have seen in that position.

Since the IF response is shot anyway, I will probably try to touch up the curve by sweeping it with my signal generator. The +/- 100KHz maximum deviation from my sig gen isn't optimal, but it is the best I can do at this point in time.

Frank N.

[oldeurope]

Originally Posted by YC-156

it doesn't have a limiter stage, and I have also now realized that the AGC is effectively disabled during FM reception! .

Hi Frank,
this is the limiter, it is called "grid-leak limiter".
g2 voltage is low and no voltage is applied to g1 like the sync separator
in a TV set.
I wanted to make a scan for you, but the pages 301 to 317 are stolen
in my book.
I know that someone stole some pages, never lend books to other people!
If you get the book please let me know.
Or does anyone else have it and can scan this pages?

Kind regards
Darius

[howard]

Hello Frank,

Gosh, a project like this makes my efforts look feeble Thats a really nice radio, congratulations on a job really well done

Howard

[YC-156]

Today I could finally put my 'official seal of approval' on this project, and put the covers back in place on the radio. A few days ago I managed to sort out the audio balance, and today I finally got around to do the last adjustments needed, that of accurately tuning the 10.7MHz IF and the FM discriminator using a sweep generator. Unfortunately my signal generator, when it is wearing its sweep generator cap, will only provide a maximum of +/- 100KHz of sweep. I decided it had to do and get on with it.

Along the way I learned several lessons:

• The tuning is quite critical, and I highly doubt I could have reached these results by using only a signal generator and the official tuning procedure from the service sheet. The IF curve shown above in graphic form was what I got for following Grundig's official procedure to the letter. In fact several of the cores needs to be adjusted so accurately that I doubt I will attempt to reseal them with wax out of fear that they will shift as I seal them.
• Perhaps not surprising, but anyway: The IF tuning also depends on the AGC level. It is well known that vari-mu valves exhibit a shift in their input capacitance at different bias points, but this was a clear demonstration to me. (Some valve datasheets, like that of the EF183, even gives sizes for the cathode compensation resistor, something which is not used in the 3033). Since I mainly listen to strong, local stations, I just told the signal generator to pour it on.
• Tuning also shifts with the temperature inside the radio. I discovered this by taking a long break in the afternoon. When I came back and powered up the set again, the nice curve I had left it with was a bit out of whack. So I let the set get thoroughly warm before proceeding.

And here are the screenshots of my results for the day. The first one shows the IF passband after quite a long tuning session. Compare it to the 'before' curve with a similar Y scale. There are seven 10.7MHz IF tuned circuits and this is the best I can do I believe. The asymmetry comes about because the VHF tuner contains two dissimilar 10.7MHz resonators. If you want a symmetric IF response, you can have a maximum of one 'odd' tuned circuit, usually the FM discriminator, the rest needs to be paired for symmetry.

http://www.warpspace.com/glowbugs/oz...assband_sm.jpg

(High resolution version.)

Horiz: 10,7MHz center - 20KHz/div.
Vert: Arbitrary Y axis (AGC response), IF response inverted since the FM detector uses 'negative' logic.

The scope trace is a bit fuzzy, as the switched mode power brick for the digicam managed to cause interference.

The discriminator:

http://www.warpspace.com/glowbugs/oz...minator_sm.jpg

(High resolution version.)

Horiz: 10,7MHz center - 20KHz/div.
Vert: Audio output voltage inverted.

After having listened to the set for a few hours I'm quite pleased with the overall result, especially the final IF tuning really brought the distortion down. Harsh syllabels are all but gone now, and instrumental music sounds great, assuming the station I listen to have a competent audio technician.

Thanks all for your help, advice and patience with this little protracted project of mine. Now on to the next one.

Best regards

Frank N.