Making a Four Vacuum Tube Radio from scratch - need help with distortion

[Daniel Bingamon]

I'm making this radio from scratch, I'm close to completion but I have a little distortion problem.
My design I call the AS4 - for American Safe Four since it has Transformers for the Heaters and the B+ as well as an Audio transformer.
It has four vacuum tubes, instead of a tube rectifier I'm using a full wave diode bridge.
The B+ transformers is a 1:1 ratio of American mains, loaded it has about 171VDC across the first filter capacitor set.
My outdoor antenna is two slinkies (yes, the metal toy) stretch out in parallel on top my workshop.
There is a 50,000W AM radio station, WLW less than 5 miles away by air - I'm a little concerned about that getting into the set.
The chassis is an old small computer that had two drive bays.  The speaker is on one bay and the variable capacitor is on the other.

A photo of the unit is attached.

And here is a video of some testing with a station playing in the background.

My questions:
1. What could be causing the negative going spikes that are stimulated by increased amplitude?
2. Is a 1M or 500K a better choice for a volume control pot?

[SuzyC]

Your problem might be the output screen grid resistor 100k ..should be around 2.2K and with a bypass cap. (20uF)

Should also have .01-.1uf bypass caps near each plate supply to I.F. transformers and maybe a decoupling resistor(1k) between mixer plate supply and IF stages power supply.

C8 is also a rather large capacitance for its intended purpose. L1 should have a tap around 20uH to couple to the slinky ant and this would greatly increase your tuning selectivity. This is needed when nearby to Hi Power AM station.
500K is fine for volume control.

Where's the picture?

[Circlotron]

Your problem might be the output screen grid resistor 100k ..should be around 2.2K and with a bypass cap. (20uF)

Given the lowish 171 VDC HT supply I'd say connect the screen directly to HT without any resistor. Quite normal practice. 100K is way way high for a power output stage.

[Daniel Bingamon]

The screen measures about 101VDC.  Is that too low?

[T3sl4co1l]

That's probably about right at DC, but without a bypass cap it'll completely tank as signal current rises.  Even with a bypass cap, it'll heavily compress the audio (including distortion from the reduced maximum power output under that condition).

You might couple the detector audio output to another amplifier to see if it's linear at that point, in which case that implicates the output.

Speaking of maximum power and matching, what is that OPT anyway?  This?  https://www.parts-express.com/70V-10W-Line-Matching-Transformer-300-040 The wire colors don't seem to match up at all, I'm not too sure.  These also aren't made for DC, it's not obvious how much they can really handle before distortion sets in.

Tim

[Daniel Bingamon]

I tried the screen directly to 1st B+ or to B+ after R8 1.2K causes a loud high pitch squeal that changes pitch and tone with volume adjustment.

That OPT is indeed the hyperlink that you referenced on the yellow tap on the primary which should make the impedance around 1960 Ohms and 8 Ohms on the secondary.  You have a good point, when I first designed it I was thinking about 120VDC not aware what the B+ would wind up around 171V.  The load resistance on the 12L6 is 2000 ohms at 110V (and 4000 ohm at 200V) which probably would be okay with 1960 for 110 but since it's coming out closer to the 200V spec which is 4000 ohm, I can try the white tap which is 3960 Ohms but then it will be marginal on the Wattage.

[Daniel Bingamon]

I tried the other tap on the OPT, no change in behavior.
This time I used the audio input instead of the radio with a tone generator.
I can still touch the screen with a multimeter and the volume increase to a nice level. 
But so far, hanging capacitors on there however cuts out the audio, I'm beginning to wonder if the tube has issues -- even through it supposed to be new.

Your comment about audio linearity is interesting, there might be an issue there as well, the volume doesn't climb much until you get to the max and then it changes a lot.  I might have wrong pot on there.

[Circlotron]

That cap from the output tube anode to cathode to roll off HF, while technically correct, is relatively uncommon. More usual to connect it straight across the output transformer primary. The way it is now, if the cathode bypass cap is marginal at HF then the cap from anode to cathode is a positive feedback path so it may well oscillate.

[Daniel Bingamon]

That could be possible. I'll try that.

[T3sl4co1l]

Oh, yeah, I bet the poor thing is oscillating.  Stick a scope on it.

The cap might be making it into some kind of Colpitts oscillator or the like.  Almost any spooky conditional behavior can be blamed on oscillation, at least until it's been eliminated as a cause.

I think I would be surprised if that tube oscillates much beyond 100 or 200MHz, which implies resonant circuits on the order of that, tops.  Which in turn implies wiring lead lengths on the order of 10s of cm, give or take circuit and tube capacitances.

The best mitigation is a "grid damper", a series resistor to the grid, typically 100 ohms or more.  This degenerates the grounded-grid amplification mode that is probably at work.  Similarly a resistor or ferrite bead on screen and cathode may be helpful, in case the grid is still getting enough feedback at some frequency.

Other possibilities could be gas in the tube (may or may not be visible through the glass), or motorboating or interstage feedback (likely you already established that's not it, after inducing it with the alternate screen supply test!).

Tim

[Daniel Bingamon]

Wow, you were right!  I took the C3 capacitor out and put it across the OPT instead.  Now the signal is now crystal clear, no distortion even at full volume.   It was oscillating, I think my touching the screen with the multimeter probe was somehow interrupting the oscillation. 
The 50KW radio station 3 miles from here is giving me a few problems as well.  The volume control leads might be acting as an antenna.

I'm also going to revisit the homemade remaking of the IF transformer but at least now I'm not chasing as many problems.  But building a set totally from scratch does teach you a lot of things.

[Circlotron]

The best mitigation is a "grid damper", a series resistor to the grid, typically 100 ohms or more.

Yep, sounds like a plan. 

https://www.aikenamps.com/index.php/grid-resistors-why-are-they-used

http://www.valvewizard.co.uk/gridstopper.html

[Circlotron]

Photos please!

[Daniel Bingamon]

Here is the latest schematic:

[Circlotron]

I mean a pic of the actual radio!

[Daniel Bingamon]

I will, sometime tomorrow evening.

[Daniel Bingamon]

Some of you asked for photos, here they are:

[Daniel Bingamon]

The new problem is that the Radio is receiving the local 50KW station 3 miles away even when the 12SA7 Mixer tube is physically pulled out.

[Daniel Bingamon]

Is it possible that the signal is traveling on the AGC linee through the IF1 Transformer and into the grid of the 12SK7?

The station is exhibiting wide bandwidth like a crystal doesn.

[wizard69]

Really nice work!   That radio deserves a finely crafted wood cabinet like we would have seen in the 1930's.

By the way I have to thank you for the post, I never learned much about tubes in tech school.   Literally had one instructor saying you don't need to worry about them.  So posts like this are very informative and educational for me.

[Daniel Bingamon]

If you're making Guitar amps, repair heirlooms, receivers from your childhood.   Or my favorite reason - because it's there.

[T3sl4co1l]

Oh yuck -- that circuit layout and the long wires going everywhere, are probably going to, well,

The new problem is that the Radio is receiving the local 50KW station 3 miles away even when the 12SA7 Mixer tube is physically pulled out.

...Yeah, that sounds about right.

You can try an EMI filter on the mains, so that the antenna connection is the only way in (and do something about the phono in, and, why is there a receptacle in the corner..?!), but the fact that there's so much coupling between all those circuits basically means there's no point in cans over the IF coils either -- and they put them there for a reason -- instead it's a big mess of coupling between circuits that need to be isolated to get the dynamic range and image rejection required for a proper radio receiver.

Tim

[Daniel Bingamon]

Tim,
Thanks for the comments.
I'm working on a revised PCB do deal with coupling issues.   Having JLCPCB available to make prototypes has been such an opening for me to make various things - for it's a lot easier than breadboarding because for one thing I'm blind in one eye and with age I need a lot of light when I do things and I can do this part on the computer screen.
How is this any different then the jillions of wires spewing about in an AA5?   I know, everything is closer together in many of them.   Also, at one point I had shield wires on the IF and the VC but something was wrong and I removed it.   I will put it back but this time I'm going to do it sequentially instead of changing too many things.

The signal isn't coming in from the mains, it's coming from my series of two stretched "Slinky" antenna at the apex of my garage roof outside.   I will probably have to make a trap on the antenna.

At one point it was receiving other stations, but something has changed or went out.
I wonder if the distortion was influenced by a heterodyne of this signal that was bypassing the IF and being out of phase it might have been fighting the normal signal that comes through the mixer and IF.
I ordered a set of IF transformers from Surplus Sales of Nebraska. https://www.surplussales.com/INDUCTORS/pdf/ti-13-pc1.pdf
I really don't like the slug tuned type with top and bottom entrances, I prefer the old dual capacitors on top style but if it work I'll be fine with it - it lends more to the hobbyist nature of this set.
They're NOS so I'll have to inspect them carefully.

Someone asked why I didn't make a nice old style cabinet.  The nature of this set is that I designed this from a culmination of various AA5 schematics, located all the various parts, calculated and hand wound the antenna coil and later remade it on a PCB board which I also designed all from scratch.   Building this self-teaches me various things which most people have learned in a classroom.  I'm not a student of any school except the hard knocks of living.    I've been tinkering with electronics since 5th grade and I'm 59 years old.
I've been an SWL'er since childhood and got into Ham Radio in my 20's.  My day job is computer programming in a data acquisition environment (which calls for a certain amount of electronics - I also design interface boards at work).   Then I got into making Tin whistles as hobby/business for about 20 years and I did less with Electronics during that time except for a lot of Audio frequency calculations for the toneholes.   
I built a 3D printer from scratch with the Reprap, Prusa I2 design as the baseline.
And with the past two years the electronic bug has come back and I've built some Amplifiers, a preamp and I wanted to get a better understanding of tubes and there is a lot more stuff and materials available in Radio to do just that.
This has been my own "self-taught" way of learning things, I give my a project and then I work on it.   I also write software at home to keep the creative juices flowing and I keep ahead that way.

[Daniel Bingamon]

Photos of the completed AS4 Radio.

The 2nd stage IF transformer was backwards.  Once corrected it was able to align properly.
Thanks to everyone here for their help.
The set works similarly to other sets of the same types of tubes.  Being a metal case it does require an external antenna.  I might design a plate in the back slightly offset to put a coil type antenna in place.

Designing from various schematics, drawing the schematic on the computer and creating vacuum tube libraries for Tiny Cad, acquiring parts, designing first a wire-to-wire version using relay sockets and a DIN rail, then designing a PC board and troubleshooting.   Quite an educational experience.  Project time from November 2020 to September 2021 - with gaps in time span.