Log of the Collins 30K-4 Restoration Project - Greg Buchwald
Sometime in the mid-1990's: Visited ENWCo. / Elgin Planetarium; saw the Collins 30K4 having just been saved from a dumpster trip by an unknowing soul that thought it was just "clutter", met Gary Kutina, and discussed the possibility of restoring the transmitter. Gary handed me the manual for the transmitter and I took inventory of missing items (tubes, etc). The project was brought up from time to time in emails between Gary and I but we were both occupied with other more pressing activities and duties.
September, 2007. After returning on a red-eye from an IEEE meeting in Kona, I grabbed 20 winks, got up, and drove over to the Planetarium for a private tour with the Northwest Suburban Astronomers. I missed the first part of the tour, which included a visit to the site of the ENWCo. factory and the trackside rail station which is now awaiting reconstruction and restoration but did see the sky show in the planetarium and the tour of the dome that contains the transit telescope. Gary and I began to renew discussions of restoring the Collins which was still sitting in the very position that I had left it in a decade earlier. It was time.....
I spent several months gathering additional spare parts including modulator tubes, all audio driver tubes (which were removed years earlier - the transmitter had never been used for voice / phone operation), power supply capacitors, a new 4-125A final, a "cone heater" for the power reduction / "tune" setting (someone must have needed it to keep their feet warm), crystals for 3885kcs and 7290kcs (the 3885kHz crystal will be replaced with 3880kcs so as to conform the the monthly Collins AM net frequency), a spare 6V6 and 807, a pair of 5R4's courtesy of Mark, K9AM, and a pair of 866A rectifiers. Good thing - the 6V6 was quite weak, the 807 was strong but it is nice to have a spare, one 866A was bad (too much internal arcing by someone that did not know how to properly "season" a mercury vapor rectifier tube), the 5R4s were on the low side of good, and the 4-125A looked well-discolored. Little did I know just how much more this rig would need....
January 31, 2009: Gary and I met at 10AM and had our first hard look at the transmitter. Tubes were replaced and the low voltage circuits were slowly brought up on a variac to insure that no damage would result from a bad capacitor or shorted transformer. After about 1 hour, the AC line voltage had been raised to 115VAC - no smoke had escaped and the bias and low voltage supplies all seemed good. When switched to the "phone" operation position, the 75TH filaments (actually, now 100TH tubes) lit as they should. One 75TH had an intermittent filament and indicated a filament to grid short. The 75THs were removed and replaced with 100TH tubes. Since 75TH tubes are more scarce than 100TH tubes, it was decided to re-bias the modulator so that the 100TH could be used semi-permanently. That modification is yet to be done. Things seemed to be going well except we could not develop any 807 grid current. I had used my allotted time and would have to put that off until next time.
February 7, 2009: Ready to begin anew, Mark (K9AM) joined Gary and I to work on the transmitter. It was determined that the PA deck would need to be removed. Besides, so much dust had collected on the chassis that the 807 grid current meter could not be read. Fifty-plus years is a long time for dust to collect on anything! Once we began looking in the 6V6 crystal oscillator section, it had become apparent that evil things had been done to the transmitter in barbaric fashion. Numerous modifications had been done to the stage in, what looked to be, an attempt at interfacing an unknown VFO to the transmitter. This would have not been a problem if they had simply injected it into either a crystal port or by removing the 6V6 and driving the 807 directly through the tube socket. But, that would have been too easy..... I might add that there is an amateur radio callsign on the door leading to the transmitter room: WB9DUJ. Mark learned that the callsign had been the possession of a club entitled the "Elgin Planetarium Radio Club" that existed during the mid-1970's. Nothing further is known about the group; information about this group would be greatly appreciated.
Upon removing the deck, it was obvious that several modifications had been made. At first, we found the 6V6 crystal oscillator cathode choke to be open. I carefully removed the choke and found a break in the wire near the connection lug at the chassis end of the choke. I was able to carefully repair that. The choke was re-installed and the deck was reloaded into the cabinet. Upon applying power and keying the transmitter, no signal was generated. Drat! The deck was once again removed and I dug deeper into the 6V6 circuitry. I found several disc capacitors connected to the tube socket. That surprised me since postage stamp silver micas would have been typical for Collins equipment of that era. Upon further investigation, I found that the original caps were still in place, but their leads had been cut right up against their bodies. The disc caps replaced them but they were of unknown quality. I decided to order new silver mica caps...at about $1 / each....from Digikey. I have an extensive junkbox at home, but I figured I'd use new parts, get the transmitter running properly, then try to replace components with historically-correct types and styles. I understand that the Digikey silver mica caps are not the original style, but they are resilient and, since the originals were trashed anyway, I had little choice if we were to get this thing up and running in a timely fashion. When we last tried the deck, we noticed smoke rising from the region around the 6V6. Not a great sign, but, in retrospect, also not bad - we knew power was getting there and, since we found the cathode resistor to be the smoked culprit, it didn't worry us too much. We could go no further with this section as I was already running late, so I spent the remaining time checking out all the power supplies and control circuits.
February 13, 2009. Mark and I head over, with Mark's harmonic, David, in tow. The new caps had arrived from Digikey along with some resistors such as the 470 Ohm 6V6 cathode resistor. Upon closer investigation we found, "that ain't no 6V6"; the faded marking on the side of that fire bottle looked like 6V6 but the tube didn't look right to me. It had dual plates and the way it acted by smoking the cathode resistor after the cathode choke had been fixed, did not make sense to me. This looked like a 5Y3GT or other power supply rectifier tube. A very close look at the tube revealed it was....a 5Y3. Wrong tube in the socket for 20+ years? Guess we'll never know for sure. But, the tube really did become a diode form B+ to ground and quickly let all of its smoke out of the cathode resistor. At least it was the sacrificial element; not the 4 layer, pi-wound cathode choke. Fortunately, I had ordered spares of the 47k grid resistor and 470 ohm cathode resistor. It was decided that the 470 Ohm cathode resistor had to be replaced. Once out of the unit, a big crack was found on the resistor. We did the necessary surgery and the deck was once again re-installed. At the same time, we found that the contacts on the keying relay had been "adjusted". The keying relay applies plate voltage to the 6V6 and 807 stages. It is capable of 75WPM CW keying speeds. Furthermore, upon resetting the contacts to a proper spacing, we found that the contact resistance was quite high. Hoping that this expensive, very hard to find high impedance keying relay was still good, we carefully cleaned up the contacts and a good, low resistance would obtained. The grunge taken from those contacts was horrific. Seemed like someone thought 80/90 weight gear oil would make a good lubricant for Morse keying relay contacts. The 50+ year old bottle of "grease" was found on a shelf, nearby. One more try....but no good; still no 807 grid current. Time to leave once again - it was late on Friday night. Tails between our legs, out the door we went.
February 22, 2009: Mark and I arrived at 1:30PM. We immediately took voltage measurements at the 6V6 socket. Where we once had screen voltage on the 6V6, we now had nothing. Once again, the deck was removed and the 20k / 10 watt resistor was checked. Ah haaaa - intermittent. I had a few resistors that could be strung in series, hidden under the deck, to temporarily repair the damaged part. It had clearly been mechanically damaged. At least it was not from us. The deck was reinstalled for what seemed to be the 30th time and the unit was fired up. Still no 807 grid current BUT a weak signal could be heard on 3885kHz! The oscillator was running. Next, a chunk of B/W miniductor was cobbled into a makeshift plug-in coil to resonate the grid of the 807. Once that was installed, the deck was fired up and the oscillator tuning capacitor was given a spin. Success - grid current. About 6.5mA worth - perfect! Next, a plate coil was installed for the 807 and it was also resonated. This time, we got nearly 7mA of PA grid current. Low, but still there. A second set of coils were rigged up for 7290kcs to occupy Frequency / Channel 2 locations in the grid and plate circuits. Real success - about 7mA of 807 grid current, but we were also able to get over 25mA of grid current - the grid current meter pegged to the right. The 807 plate tank plug-ins have provision to add series resistors in the 807 cathode circuit to adjust drive by degenerating this stage. It seems the 75M coil was set for a high cathode resistance but the 40m coil was not. Since we promised Gary it would be an earlier day and it was now 4:30, we decided to call it quits. I took the 807 grid and plate coils home so that they can be permanently installed in their respective coil cans. Since we have already tested the HV power supply on February 13, I am confident that we will get lots of RF out of the transmitter next time and we will be able to move on to the modulator.
February 24, 2009: The lack of 807 grid current without a 6V6 plate tank in place bothered me. The manual indicates that no oscillator coils are needed below 6MHz and that dummy coils (empty coil forms and shields) were supplied to allow the retention bracket to work properly. Since the original frequency was 4.975MHz, there were, indeed, dummy coils in place. Theoretically, I would need a coil for 40M but 75M should work without it. The manual also indicates that, for operation below 6MHz, excessive PA drive could be compensated for by increasing the capacitance of the 6V6 plate tuning capacitor - it was simply a capacitive divider shunting RF from the 807 grid circuit if a resonating coil was not in place. Finally, I noticed that the "typical meter readings" section showed an 807 grid current of 0 - 7mA. Everywhere else it speaks of grid current, but, in fact, under certain conditions, the 807 is operated Class AB1; no grid current is drawn. Mystery solved; it was perfectly normal to show no grid current below 6MHz. I decided, however, to actually wind coils for 75 and 40M and draw 807 grid current on each band, thus running the 807 more efficiently.
Another oddity was also noticed: We had noticed that a small neon lamp, mounted behind an amber jewel bezel on the rear of the PA / RF deck, would occasionally light or flicker. Initially, we could not figure out what the lamp was for since there was no reference to it in the manual or on the drawings / photographs. Thinking the neon bulb was old and in need of replacement, we put it on the list of items to find. Upon looking at the parts list, I found one obscure reference to the lamp as a transient suppressor. The lamp is actually across the coil of the keying relay, which operates at about -55 volts from the bias supply under normal conditions. This is below the firing voltage of a neon lamp, so it must remain unlit when static. When the relay is keyed, however, the voltage soars due to the inductive "kick" in voltage as excitation is removed from the coil and the field collapses. Today, we would put a snubber diode across the coil but in the time of the transmitter design, selenium rectifiers were the only non-vacuum option and they were not tolerant of such spikes. Rather, the neon lamp was placed across the coil and, when the solenoid filed collapsed, the bulb would fire and limit the peak voltage to about 65 volts - pretty smart design!
March 25, 2009: Since I had decided to replace the 75TH modulators with 100TH tubes, the grid bias for proper Class B operation of the new tubes would have to be changed. The 30K-4 has a potentiometer for adjustment of the bias; however, the range is not adequate to allow for proper use of the 100TH. With 2000 Volts applied, the 75TH requires about 90 volts of bias on the grids for proper operation whereas the 100TH requires only 35 volts. I decided to re-do the bias voltage divider resistor network to accommodate the revised voltage range for the new modulators. A couple of new, fixed power resistors were ordered from Digikey as well as a new 25 watt Ohmite wirewound potentiometer. Amazingly enough, an exact pot was is still manufactured today and Digikey had it in stock! Don't ask about the price.....
April 11, 2009: I met Gary at 8:30AM to work on the PA stage. The oscilator stage was temporarily tuned using a peice of miniductor coil stock connected to the dummy coil. Since the coil stock did not have enough turns, a small mica capacitor was added in parallel to achieve resonance. The dummy coil was used as base so it was easily installed. Tests indicated 4mA of 807 buffer grid current - a good sign. Next, the 4.5 - 8MHz stock buffer plate coil was brought into resonance by adding a 55pF (qty 2 110pf silv er mica 500V) capacitor accross the coil. Over 25mA of 4-125A grid current was immediately achieved - another good sign. To reduce the 4-125 grid current, 1500 out of 2500 Ohms of cathode degeneration resistance in the buffer stage was added into circuit. Ideally, about 15mA of PA grid current should be observed; at resonance of the 807 buffer and peak tune of the oscillator stage, about 23mA was observed. This was easily brought down to 15 - 20mA by slightly detuning the oscillator stage.
With proper grid drive avaiable, it was decided to try to get power out of the tranmsitter. The transmitter was switched into the "test" mode whihc places a 660W heater element in series with the 2500V plate supply. In addition, the plate transformer was strapped for 2000VAC either side of center tap vs the normal 2500 volts. Upon brining up the HV, about 20mA of plate current was observed; far short of what was expected. I then switched the unit to the operate position at which time it began to cycle and the 4-125A filament fuse opened. The fuse was replaced (2A) and the unit was once again brought up in the test position. This time, 200mA of plate current was observed and the 4-125 plate began to glow a dull orange color. The stage could not be broght into resonsance; the plate coil is too small being proper for 4.5 - 6MHz vs the 3885kHz we have been tuning up to operate on. But, at maximum capacitance, about 1/2 amp of RF line curernt was observed: That amounts to 12.5 watts into the 50 ohm load.
I am waiting for US Customs to release my 30k5 from New Zealand. On the PA deck for that rig, are 2 coils for 3 - 4.5MHz. I will trade one for the Elgin 30k4 - that will giev them the ability to operate on 3885 / 3880kHz as well as 7290kHz; I'll have the same ability.
The new 6B4 was also fitted and yet another oncorrect tube was also found to be placed in the audio deck. A 6SN7 was incorrectly replaced with a 6SJ7. The proper 6SN7 was fitted to the transmitter and the spare 6SJ7 was set aside for future use, if needed.
All told, the RF deck had to be removed - we traced out the antenna relay panel. Documentation for Collins equipment is usually very good but the relay panel is barely documented in the manual. I will post a copy of the schematic along with anotated photos of the panel - that should help others in the future that are trying to bring these old rigs back up to speed.
Time was up; I had to pack up. Next time, we will rebuild the bias network for the modulators as described above as well as, hopefully, getting full power out of the transmitter. We are still trying to get the transmitter operational for use on the Collins net by May.
April 21, 2009: My 30K5 has finally arrived from New Zealand and has been cleared through customs. I picked it up at the bonded freight yard in Addison, IL and tgransported it to my WI hamshack. The much-needed coil for the PA of the 30K4 in Elgin arrived intact. I plan to take the coil over on May 1 and install it into the Elgin 30K4.
May 1, 2009: A special event was held at the Planetarium this evening to kick off the countdown to the 100th annivesary of the ENWCo Observatory. Gary Kutina asked me if I would attend the event and ask questions about the 30K-4. Of course I agreed to be there! Mark, K9AM, attended the event with me. I arrived about 7PM and started by answering a few questions. We fired up the box for a few photo ops with the 4-125 blazing away. After the photos were taken, I installed the new 75M PA coil that I got from New Zealand. After a few iterations, I was able to get over 100 Watts into the dummy load while in the "Tune" position. In fact, I could not get the plate current at resonance low enough. We then tried to switch tot eh "operate" position. This resulted in a rapid chatter of the keying relay for a few secondsfollowed by popping the circuit breaker; the latter not only silencing the transmitter, but also killing many of the lights on the first floor where many of the guests were located! The breaker panel was located and the proper circuit was reset. Excitement over, we decided to try to figure out what, exactly, was happening.
I then used the G3SEK Excel spreadsheet to determine what the Pi Net values needed to be. I quickly learned that the output plate load, assuming Class C operation with a "k" of 2, needed to be 8k Ohms. The problem is that transforming this to 50 Ohms requires a "Q" of at least 20 to get a realistic result (assuming 2500VDC @ 150mA). Back when the 30k4 was new, most antenna systems were either 300 / 450 Ohm balanced feeders or a direct lead from the transmitter to the antenna - effectively part of the antenna. They were generally not low impedance like 50 Ohm coax. I did match the network by allowing a Q of 24. This, in turn, required less than 50pf to resonate the plate. That meant cutting the link between the 2 sections of plate tuning variables. Once this was done, the plate current, in the tune position, could be reduced to 60mA when the loading capacitor was fully meshed. Next, we tried the "operate" position once again. The transmitter actually seemed to work normally....for a few seconds...before a big spark flew from the top of the PA deck onto the chassis. The PA blocking capcaitor had been leaking some resin; it decided to "let go." But, that was a good thing. It pointed out a bad componant, probably located the cause of our ovedrloads earlier, and did indicate the box would make full power. We decided to call it a night - I had a long drive ahead of me as I wanted to attend the Superior (WI) hamfest the following morning - a 6 hour drive - and it was already 9PM. I drove 3 hours to a relatives house, hitting the rack at 1AM and tehn getting up at 5 to make it to the fest by 9:30 (it started at 9 but everything I needed was still there).
I want to state that the Superior hamfest was great - I will drive up for the next one! I had a great time and met some really great hams, too. Plus, driving almost 400 miles north made sure that I was looking at a whole new variety of goodies from a whole new bunch of junk boxes!
As a postscript, I did attend the hamfest and was able to find a fair number of replacement caps for the blocking capacitor. I also thought about the high overall plate impedance. I think the best course of action will be to reduce the pate voltage to 2kV, or even slightly lower, using the taps that exist on the plate transformer and then running slightly more plate current. By doing this, I can get the plate impedance down to about 5800 Ohms - still high, but much more managable. I would be able to run a "Q" of 15 or so, reduce I2R losses in the PA coil, and still get close to 250 watts out of the transmitter. There would be a slight mismatch of the modulation load but nothing of any consequence. I plan to find an autotransformer, buck-boost, or 100 - 115V transformer at Dayton to drop the pate voltage to the above-mentioned level. A variac is probably the best but I need to find one that is at least 800VA.
Next visit will be just after Dayton......
August 18, 2009: After a 2.5 month hiatus from working on the 30K4 due to busy schedules on the part of Gary and myself, we inally found time to work on the old gal again. We began by removing the PA deck and replacing the 1000pf/6kV plate choke bypass capacitor that leaked out all of its resin. I had noticed a small amount of resin on the 4-125A PA tube but, during our last visit and while trying to bring the transmitter up to power, we ended the night with a spectacular bang followed by a few sparks from the region of the capacitor. The new cap was installed and the deck replaced. Next, I had suspected a problem with shorted turns on the plate transformer due to the high amount of 60Hz hum heard when the transmitter plate supply was enabled. Even in the "tune" mode, the hum was excessive. Desiring to reduce the plate voltage anyway, I borught along a transformer that would produce just over 1500VDC with 120 volts on the primary and about 1875VDC with 140 VAC on the primary. My goal was to run the transmitter at 1850VDC and 160mA to produce 200 watts output assuming 68% efficiency (which was actually confirmed at 69.5% later during the evening). A variac will be installed on the floor of the transmitter so that we can adjust the high voltage from just over 1000VDC to just under 2000VDC. As it tunred out, the secondary of the original transformer was shorted from the 2000 volt tap to the 2500 volt tap; I'm surprized it worked at all! We will look into rewinding it but, for the purists out there, the replacement transformer is actually a very rare 30k4 plate transformer that was special ordered back in the early 1950's for another transmitter that was suppossed to run at a reduced power level (150 watts phone) and get extremely long tube life. I located it several years ago as a spare for my own 30k1 but decided to use it in this application.
With the new trransformer in place, we brought the unit up slowly; once again on the variac. Everything came up normally and we decided to try it at 120VAC. After short wait for mercury in the recitifers to come to temperature, we hit the Remote / Local switch and the the Key switch. We expected to see grid current on the PA but....nothing. Then it dawned on us that we forgot to close the Plate switch! When that was done, in the LV position, the oscillator came to life. We peaked the grid current, then limited it to 15mA. Next, we switched to the Tune position and brought up the high voltage. The PA came to life and we tuned for a dip and loaded for max. We got about 0.8 Amps of RF line current into 50 Ohms with about 90mA of PA current. It was time to go for broke. Moving to the Operate position, we retuned the PA and got 1.7 Amps of RF line current which amounted to 145 watts at 150mA. Running the plate voltage up further with the external variac, the power just continued to increase! By reducing the plate voltage, we are able to get the PA load impedance down from over 10k Ohms to 5800 Ohms; still high but certainly tolerable. When we operate at our desired 1850VDC at 160mA, our output will be 200 Watts into 50 Ohms assuming 68% plate efficiency. At that point, our plate dissipation will be about 95 watts - no problem for this tube.
Happy with our results, we decided to call it a night. For a finale, we switched to Phone mode and gave the audio gain pot a twist. The modulator plate current jumped as the noisy pot was rotated. Next time: Set up the mod bias and put some audio to this rig. After that....on the air! This rig has probably never been on the air in the phone mode as the ENWCo ony sent keyed carrier. Some hams did try to get it runnign a few decades ago, but they did not seem to get it going. In fact, I think a big mismatch of the PA deck probably caused some of the damage we hve found as well as the blown plate bypass cap.
For the curious, on 75M our plate impedance will be 5800 Ohms, matched with the Pi Network operating with a Q of 15. That requires 90pf on the plate side, 17.7uH for the inductor, and 705pf on the load end.
More to come.....