Howard 474 AM/FM Mantel Radio


Left to Right, the controls are On/off volume, AM/FM and tuning. The dial is lit with a #47 dial lamp.

This set is the most well known of the commercially made Fremodyne sets which appeared in late 1946. Externally it looks like the AM only 901A, except for the addition of the extra knob for switching between AM and FM. The cabinet is either white or brown bakelite. Unless you have already done so, I suggest a look at the Fremodyne article to learn what is so unique with this receiver.

Introduction
The 474 is a typical US type of AC/DC mantel radio design with the Fremodyne FM bit tacked on. 150mA heater valves are used. For AM, the 12BE6 functions as the converter, the resulting IF being amplified by the variable mu 12BA6. Detection and audio amplification is taken care of by the 12AT6. The audio output uses a 50L6; the only octal valve in the set.
Unlike Australian sets, loop aerials were standard in US mantel radios, and the Howard is no exception. In fact when you think about it, Australia sets were about 10 years behind. In 1946, Australian sets were still using octal valves and full size components. While ferrite rod aerials did appear in the late 50's in some Aussie sets, the standard piece of wire trailing out the back functioned as the aerial right up to the end of valve radio production for most sets. It took until the mid 50's for the 7 and 9 pin valves to appear in mains operated sets here, although the 7 pin battery valves had made their presence here earlier.
Internally, the Howard would not have looked out of date in the 1960's in Australia.
For FM, a 12AT7 functions as the Fremodyne circuit. Audio output is fed into the 12AT6.
Have a look at the service manual  here

The Power Supply
Like all mantel sets produced in the US from this era onwards, a series heater circuit is used with all valve heaters wired in series across the 120V mains. Of course all heaters must have the same current rating; in this case 150mA. In the Howard, the sum of valve heater and dial lamp voltages add up to 106.7v. A negative temperature coefficient thermistor absorbs the rest of the voltage and provides heater surge protection when the radio is first turned on.
B+ comes from a half wave selenium rectifier.
Of course, while the expensive and heavy power transformer is eliminated, it means the mains is connected directly to the internal circuitry and often to the chassis as well.
This creates a safety issue and is why sets using this type of power supply were very unpopular in Australia. In fact, apart from some mains/battery portables made in the 50's, the technique was only used for a few radios made for DC mains areas which were extinct by the 1950's.
Some sets did try and reduce the shock hazard by having a separate earth busbar running around the inside of the set to which the earth returns were connected. This was isolated from the chassis by a condenser, typically of .01 to .1uF. Of course, if the condenser breaks down the chassis will be live depending on the mains polarity. Even so, the reactance of this capacitor at 50 or 60 cycles is low enough to allow the mains to be "felt" in the right situation. The aerial is likewise isolated, typically via a .01uF condenser. The Howard 474 has taken this approach, which is a good thing as the chassis is exposed to the user. However, the earth to chassis isolating condenser is .22uF. At 60 cycles this has a reactance of about 12,000 ohms which means at 120V, that 10mA can flow. Hardly safe! At least the "ground" connection is fed via a further .02uF condenser which has high enough reactance not to create a hazard.
Many US sets simply had the chassis and or it's mounting screws exposed without any such isolation. It is probably only because of the lower voltage mains in the US that electrocutions were not more widespread.
It is certainly unacceptable on 240V, where a live chassis set must have the chassis fully enclosed and the aerial connection made by more appropriate values of isolating condenser.


The back of the Howard. Top left is the FM aerial input, below which is the ground terminal. The wire on the left is the external AM aerial connection and the wire to the right is the line aerial. Note the chassis is accessable to anyone that wants to touch it. The label gives instructions on how to use the aerial connections.

The live chassis issue is not important in my instance, as such sets are run from a double wound 240-115V stepdown transformer.

My Howard 474
Whatever did we do before Ebay? I'd never have got a 474, being on the other side of the world otherwise. I'd always wanted a commercially made Fremodyne since I first learnt of the design back in the late 80's, not only for the novelty but also as a comparison to my own homemade versions. It took about nine months before one turned up. I paid about US $66 for it. That's high for a US mantel radio and I can only assume that one of the other bidders knew what was special about this radio.
The radio arrived intact and as soon as I opened it, I knew it had provided a home for a small mouse at one time.

All the surface rust is the result of mouse piss. Fortunately none of it had penetrated any of the wound components. The dial lamp wires had been replaced just before the photo was taken as the old insulation had crumbled away. In the front between the speaker and tuning gang is the 21.75Mc/s IF coil. At the rear of the gang can be seen the three turn FM oscillator coil and beehive trimmer.

The mouse damage was superficial though, with a few wires with insulation partially chewed. The main damage was the mouse piss on the chassis, having caused surface rust over most of it. The piss had also got into the trimmers on the variable condenser. Underneath the condition was good and no parts had been replaced.
My set is the Ivory version. The case is white bakelite, but the outside is painted in gloss white. It had the usually dirty knobs and the fibre back had a few cracks; obviously it was a well used radio, and as I found later I suspect most of this use was on AM.
The dial cord was broken (chewed through?) and the dial pulley seized.
So I powered up and after a long warm up I could tune a few AM stations. The stations weren't in the right positions and gain was a bit low. Nothing issued forth on FM; not even the superregenerative hiss.



Restoration

The Power Supply
First thing evident was the B+ was low; around 70V. The Fremodyne was only being fed with about 60V. Not surprising it didn't work. Either the rectifier or filter condensers were faulty. Placing a 1N4007 across the selenium rectifier confirmed the former and brought the voltages up to a bit more than specified. Switching over to FM, the B+ to the 12AT7 was still low and the drop across the 1K was too high for the few mA this stage draws. The 40uF filter condenser was rather leaky; about 2K resistance!
I replaced it with a new 47uF and the voltage was now around 120. Still no hiss from the super regenerator although it sounded like it was almost working with a few weak oscillations as I tuned the band.


Here is the underneath view, after I'd changed the leaky 40uF condenser. Note the selenium rectifier just behind the speaker. The Fremodyne circuit is under the middle control.

A selenium rectifier has a higher voltage drop than the silicon counterpart, so I had to place a resistor in series with the 1N4007 to restore the correct voltages. This resistor is also needed to reduce the surge current before the first filter condenser has charged. Valve and selenium rectifiers usually have high enough internal resistance not to require this extra resistor.
I found the optimum value was 15 ohms, and in view of the high ripple current, I used a 5W resistor which runs barely warm.
Trouble was there isn't much space around the rectifier to put the added components.
I could have removed the old rectifier to free up some space but I wanted to keep it for authenticity, so I had to cram in the 5W resistor next to it. I found a space amongst the wiring where it fitted perfectly and where the minimal heat would do no harm.

The AM Section
Now that we had full B+, it was time to get this working properly. Problem was the stations were not in the right places. The high end stations were off the dial too. This suggested the local oscillator (12BE6) was running too high. As it's only a function of inductance and capacitance, there surely couldn't be too much wrong. Either the inductance of the oscillator coil had gone down or the capacitance in the oscillator tuned circuit had also decreased. Like most sets of this era, the local oscillator coil is non adjustable, so it wasn't going to be a case of a tuning slug having moved out of position. By now I'd taken out the tuner sub chassis to work on the FM part, and in doing so I'd given the whole lot a good clean; and that included the grime under the mica trimmer caps for the AM band. It was a surprise when I'd put it all back together that the gain was up and the stations were in the right places. I really didn't see how cleaning the trimmer caps would fix the fault, but it did. After a check of alignment, the AM performance turned out to be very good. The loop aerial has very good pickup, and even with no negative feedback the sound is quite good.
I don't think the cheap and nasty reputation of these US made AC/DC mantel sets is justified if this set is anything to go by.

The FM section
This is constructed on a small sub chassis, shock mounted on the main chassis via rubber grommets. As I'd mentioned earlier, the Fremodyne detector wasn't oscillating. So, in order to check the components involved, this little chassis had to come out; there just wasn't enough access. Despite the cramped conditions and poor service access, it wasn't too difficult to remove. I had to disconnect the heater, audio, B+ and earth connections before unbolting it whereupon it lifted straight out.
Well, all the parts in the 21.75Mc/s IF super regenerator tested ok except the 10uF electro which was o/c. Interestingly, a 25uF had been fitted, so I replaced it with the same value. The parts list does specify 10uF. Before I put it back together I had a good look at the construction, so as to compare with the homemade units. One thing that was never really described in the literature concerning the Fremodyne is the construction of the RF chokes (Perco circuit excepted). The local oscillator cathode choke is just a standard enameled copper wire wound choke on a small former about the size of a 1/2W resistor. It measured 3.4uH. Note that the Meck specifies a 12uH choke and the calculated value for the Perco choke is about 13.5uH.
It is obviously not critical. I couldn't resist opening the IF transformer of course and while the 21.75Mc/s winding was as expected, I didn't expect to find the 21.75Mc/s choke to be wound on the same former. It consisted of a honeycomb winding of litz wire by the look of it. I didn't measure the value as it would have entailed disconnecting too many parts. As is pointed out in the Perco design, and confirmed in homemade units, there is no inductive relationship between the 21.75Mc/s tuned circuit and the associated RF choke. That it is mounted on the same former is just for convenience.
I noticed that the plates of the variable condenser for the FM band were of an interesting shape. I presume this was to aid tracking between aerial and RF circuits. Note that the 474 has a slightly different tuned circuit for the local oscillator than the 'official' Fremodyne. The oscillator coil and associated trimmer are in series with a 50pF condenser. Again, it's likely to be for tracking reasons they've done it this way. Even so, as I found, I doubt it was necessary as the tracking of the Howard turned out to be the same as the official or homemade versions. Anyway, replacing that o/c electro didn't do a thing.
Thinking about it the next day, it occurred that just maybe the 12AT7 was weak so when I arrived home I put in another 12AT7 and the wonderful hiss issued forth! Tuning across the band revealed it worked and sounded just like the sets I'd built. Needless to say the stereo and SCA subcarrier beats were just as evident.
The fact that the 12AT7 was weak did surprise me, for it is operating under very low power. That is until I realised that the set had probably been used for AM most of the time. This would have left the 12AT7 with a hot heater and no plate current...a recipe for cathode poisoning.


Two electros, two crumbly dial lamp wires and a weak 12AT7 so far replaced.

I reasoned thus it might be possible to reactivate it. Out came the trusty AVO valve tester and sure enough emission was really low in the triode used for the super regenerator. With 9V on the heater and no bias, I eventually got the valve drawing about 30mA on the plate. Enough I figured to burn off any cathode impurities.
And yes it did seem to work. For a while. After a couple of days I felt the performance was dropping off which was confirmed by re installing the replacement. Well, it was worth a try!

I can't get 2G0
Whilst checking the alignment of the Fremodyne circuit, I discovered that the performance was starting to drop off around 2MIX (106.5). 2SER (107.3) was barely receivable and full of modulation hum. 2G0 (107.7) wasn't even there. I initially thought there was dirt in the tuning condenser, but playing around with the trimmer confirmed it wasn't that. It really looked like the local oscillator was dropping out at the high end. Especially with the modulation hum appearing just before reception disappeared.
So, I took the Howard into work and checked with a spectrum analyser. Just as I thought!
It had to be something like a bad earth or faulty bypass cap. Well, it wasn't a bypass cap so I prodded around with a metal pair of tweezers testing the earths. Bridging the
sub chassis directly to the main chassis immediately brought up the oscillator. Not only that, the oscillator amplitude was much higher and consistent right across the band.
So I'd found it but what was really wrong? The sub chassis has a braided earth connection to the main chassis doesn't it? Yes, it does but it wasn't a good connection up around 130Mc/s (the L.O frequency at the high end). Bridging the braid straight to the chassis about half way along made it come good. Obviously the braid was just too long...a design fault in my opinion. This is a perfect example of why home constructors of VHF sets need to give so much thought to layout and not build them like a medium wave crystal set.
I simply soldered another, shorter, piece of braid between the two chassis. Problem fixed.


The old and the new. The original earth strap runs over the sub chassis mounting screw. I needed to add a shorter strap, visible just in front of the 12BE6 to get reliable local oscillator operation on FM. The paper condenser is the .02uF earth terminal isolator. Had this set been destined for operating off 115V mains with no transformer, I would have replaced it.

How could this have gone unnoticed when the radio was built? Well, unlike a normal superhet, the Fremodyne won't sound any different if there's no local oscillator activity as the super regenerative rushing sound will still be there. It's only because I knew there was a station at a certain frequency I knew something was up.
Why didn't the alignment in the factory pick it up? That's because the alignment is specified at 105Mc/s...which was receivable as normal in this set.
My guess is that this was a borderline case of design; most of the time it worked, but get a group of components with certain tolerances and it all added up to something that wasn't 100%.

Performance
After completing the VHF alignment I could give it a good test on FM. As I've said before it is the same as the homemade sets in terms of sound quality. When the conditions are right, and particularly on a station with no SCA subcarriers, it can produce very good sound. The sound is definitely much improved with the radio inside its cabinet, with the improved baffling.
I tried it on several aerials. The power line aerial worked for most of the high power stations well enough. It really did provide a reasonable signal, though the position of the mains lead did have an effect on this. I also tried a few feet of wire, which I thought was a bit better than the power line aerial, maybe because it could be moved into the ideal position. I could get some of the weaker stations. Finally I tried the outdoor aerial and simply connected the ribbon to the FM aerial and earth terminals. Obviously the signal was much stronger and more stations could be received.
Sensitivity turned out the be as specified; ie. 200uV. A 100uV signal was getting a bit noisy and below about 30uV it is just to weak to hear anything.
One thing that has become evident as I've played around with Fremodynes, is that the SCA and stereo subcarrier beat has a lot to with the signal strength. It seems that the beat problem is more evident with very strong signals than with weak.
The ideal signal strength seems to be when it is just strong enough to quieten the receiver. At this point, with careful tuning, the problem is considerably minimised or not even evident. I guess this makes sense as with a weaker signal the subcarrier beat will be weaker also.


The Howard 474 in 2020.
A routine repair to refit the dial "glass" opened the proverbial can of worms. 14 years later, I was surprised to see the severity of corrosion on the chassis. Admittedly, I had done little to clean it up during the previous "restoration" just described, but it seemed a lot worse now. Furthemore, the AM section was not working at all. This was narrowed down to the IF amplifier, and in particular there were no volts across the 270R cathode resistor. This was despite full plate and screen voltage. Evidently the 12BA6 was not drawing any current, and replacing it confirmed it was faulty. It would more than likely be an open circuit connection to the cathode inside the valve.


Note the corrosion on the 2nd IF transformer can, and the corner of the chassis near the 50L6.

There was no option but to do a thorough clean. This entailed removing the speaker and tuner sub-chassis. I went over the chassis and IF cans with a Dremel grinder and wire brush and this did a pretty good job of getting the worst of the rust off.


Tuner sub-chassis. Beehive trimmer is for the FM local oscillator. Note the 2pF coupling condenser between aerial and oscillator sections of the tuning condenser.


So far, rust has been ground off around the 50L6 socket and under the speaker.

I had contemplated painting the chassis, but it looked good enough now to probably not require it. And so, the speaker and tuner were reassembled on the chassis.


Chassis after further rust removal.

In the intervening 14 years, the paper condensers were growing some kind of fungus and at this point decided to do a proper restoration, replacing the rest of the paper condensers except the 12BA6 cathode bypass, and the .22uF chassis isolating condenser.
I also took the opportunity to tidy up the rectifier - removing the old selenium rectifier, and installing a four lug tagstrip as per my 3rd Howard 474. On this tagstrip is a 22R anti surge resistor, a 1N4007 diode, and a .001uF condenser to prevent modulation hum.

Upon reassembly, the FM section didn't seem to be working as well as it should. The local oscillator activity was weak, and dropped out in certain parts of the band. As previously described, this was evident when I first got the set, and was fixed by adding an extra earth braid between the chassis and tuner sub-chassis. The problem was back again, and at a guess was caused by some connection between the two being less effective now. I also considered the 12AT7 was faulty in some way so tried another - this provided no local oscillator activity. It was clearly a repeat of the design fault which appeared also in the Olympic. The trouble is the Howard and Olympic operate their local oscillator on the high side. It's harder to get things to oscillate at higher frequencies than lower, which only shows up any weaknesses. And that weakness is the circuit around the local oscillator. Instead of the 12AT7 grid receiving the full voltage from the oscillator coil, it is reduced by the voltage divider action of the 50pF padder, and the tuning condenser (assumed to be about 15pF).
The circuit is fussy about valves, as was found with the Olympic, and now the Howard. A third braid from the tuner sub-chassis fixed the problem.
It was also found that different valves would affect the super-regeneration performance. The quench frequency varies somewhat depending on the actual valve. If the quench frequency is too high, sensitivity is reduced. If it's too low, intermodulation distortion becomes problematic, as does the beat with the stereo pilot tone. The optimum frequency is around 35Kc/s. At one stage, having replaced the 150k resistor (measuring 166k), and the .0025uF with a .0022uF, it was up around 70Kc/s with a couple of different 12AT7's. Using .0033uF brought it down to 32Kc/s which worked well, but the intermodulation distortion was annoying. I put back the original .0025uF which produced a quench of 40Kc/s. This worked well enough so left it as is.

The final thing was to incorporate the anti cathode poisoning circuit which has been described in the 3rd Howard 474 article.



The Fremodyne Page

Home