project came about because I have always wanted to
build an all metal tube (such as 6SK7,6SQ7,6SA7,etc.) superheterodyne
broadcast receiver. This circuit comes
close, except for the audio output tube, which is glass.
many projects I do, I went
overboard on the
design. For example, it includes two
high-gain IF stages plus two RF stages with three tracking tuned
which are practically unheard of in AM broadcast receiver design.
Before I continue, I want to emphasize that this is an experimental receiver and, in the end, it proved to be impractical. It's construction is difficult due to the circuit complexity and the small chassis that was used. Also, parts of the circuit are hard to align and keep stable. Therefore, I highly recommend that no one attempt to reproduce what I present here. This project is for reference only.
The schematic is very large, so I have divided it into two sections, Part A and Part B.
Most of the parts are common and straightforward except for the coils and main tuning capacitor. The three RF coils in the front end are old NOS Miller (44-RF) air-core units I had in the junk box. The IF transformers are common 455 kc. units salvaged from AA5 radios. I just removed the insides from the small aluminum cans and reinstalled them into larger ones that I had. I did this to make the IF cans match the era of the tubes I was using. The local oscillator coil is homemade by winding enough turns on a 1/4 inch slug-tuned coil form to get about 120 to 200 microhenries, when tuned through it's range. The tap is about 32 turns from the ground end.
The main, four-section 500 pf. tuning capacitor was also a member of my junk box for several years. I have been waiting for a project like this, to put it to use. I still can’t remember where I obtained this part, but it is very nicely constructed. The large dial pulley wheel gives it a 12 to 1 tuning ratio. It would probably be very hard to obtain a four-section unit like this. Although it appears to be straight-line variable, the finished dial turned out to be a little crowded at the top end, more than I like.
The completed, drilled chassis waiting to mount the tube sockets and IF transformers next. I have already mounted the on/off sw. and volume control which are located under the chassis on an "L" bracket. I didn't have the correct size Greenlee chassis punch, so I used a set of regular hole saws on all of the larger holes. I wanted to use an "HP" type of socket for the AC which resulted in a lot of hand filing. The chassis is a Bud type which comes with a shiny finish and some scratches that I don't really care for. I always go over these with 0000 steel wool after drilling. This gives them a nice satin finish.
In referring to part B of the schematic, notice that the local oscillator is a 6SA7 pentagrid converter. This is an unusual choice. I had seen this circuit on a Japanese website and I always wanted to try it. I was originally going to use a 6J5 triode, but after trying the 6SA7, I decided to use it instead because it was much more stable and I like using the same tube in more than one circuit of the same radio, if I can. I decided to use solid-state rectifiers instead of a tube because I was short of space and eliminating the high-current rectifier filament eased up the strain on the power transformer. Notice I used an in-rush current limiting resistor ahead of the filter capacitors. This is necessary because of the large values of these capacitors. I used the leftover 5 volt winding to provide a –1.2 volt bias supply for the 6SQ7. This improved the fidelity of the first audio amplifier.
My goal was to use all metal tubes, but I couldn’t find a suitable one to use as the audio output. So, I chose the 6K6 operating in class A. This provides plenty of good quality audio.
Rear view showing the antenna and ground connections. Also, note a 1/4 inch phone jack for the speaker. As you can see, the main tuning capacitor dominates the chassis.
Front view showing the large slide-rule dial. I believe that this is one of the best types of dial you can use. It makes tuning simple and it allows for an accurate dial scale. I still need to make the scale which is fairly easy. Just markup a blank paper dial, transfer it to a drawing program such as Corel Draw, print it then glue it on. Since the dial turned out to be somewhat crowded at the high end, I might try to find another, more straight-line capacitor before making the scale. However, I doubt if I will ever locate one.
Bottom of chassis reveals how tight the components are arranged. The circuit board (center right) holds the large main filter capacitors. The large can at the top houses the oscillator coil and the can at the bottom contains the AC line filter. The audio output transformer is seen in the bottom left.
Although this was a difficult project, I did manage to get it stabilized and aligned pretty close. With three tuned RF stages, tracking is the biggest problem. The main tuning capacitor has cut-plate rotors which allowed me to adjust the tracking to within + or – 5 picofarad. The Miller coils were very close in performance which also helped. Another problem is the tremendous overall gain you get with so many stages of amplification. Here, I used large emitter resistors and low screen voltages along with a large amount of AGC to bring it under control. Also, part placement was critical for stabilization. The finished receiver is very sensitive and selective. However, there are a few things I would do different on the next tube superhet:
1. Three tuned RF stages are really more than is necessary. I could have used two and increased the gain to get about the same results. This would make the procurement of the main tuning capacitor much easier, and make it much more stable.
2. The design uses an outside antenna and ground connections. These work very well, but I like a built-in antenna much better, like a ferrite bar.
3. The low end of the dial seemed to be less sensitive than the middle and upper end. It appears this is due to the air-core performance of the Miller coils. Adding a small capacitor across these coils improved the low end but interfered with the selectivity at the high end. A different set of coils, perhaps powered iron core, would probably help.
4. I would use a variable capacitor that was truer to the straight-line curve so the high end of the dial wouldn’t be so crowded. As I stated earlier, I wouldn’t recommend that anyone try to reproduce this set, as built. However, the individual circuits could be used as building blocks or references to this type of design.