Finally have got the project to a running stage as a working amplifier.
This thing has turned into a running lesson on valve design.
It is nowhere near "finished" but at least a stable design to tinker with and fine tune.

I guess this pdf is a "part three" of the series Brad is attaching to post 1 in this topic.
I do not mind any comments or suggestions on the design, half the time I don't have a clue myself.
Even if the thing is is a failure in the primary purpose, its a hell of a RF front end for a AM receiver!!

I'll send the usual PDF to Brad and you can all have a laugh as I wade my way through the design.

Fred.

Part 3 uploaded.

AWV valves too - I am impressed!

RF chokes inline with each 6EH7 plate supply.

Nice work Fred.

Yes, BBTV, the natural frequency response of the assembly is a HF roll off from input to output.
That is from all the capacitances inherent in the components.
The divided plate loads gives a tailored HF lift.

Subsequent to the circuit you see, I have inserted 200p series capacitors at each grid to roll off the lower end and changed a few values in the plates.
That has given me an almost flat response from 100kc to 2000kc with a slope either side.
The amp is now much calmer with the sub frequencies rolled off.

I am getting an overall gain of x10,000 without instability so am now working in the milli/micro volt range for inputs to give a deflection on a 0-10 volt meter. I am also working in the noise floor of the valves!

I really need to get x100,000 gain and reduce the noise floor by 1/10.
Its just a matter of chipping away at each problem and incremental improvements.

Ultimately I want to be able to resolve a 1micro volt to 100 microvolt range RF signal clearly on a moving coil meter.
Lots to learn about valves still!!!!

Fred.

Servicing of several hundred radios does teach you where things commonly go awry. Early TRF's, Autodynes, & The latter Superheterodynes with Pentagrids & tubes with diodes within for detectors all have their idiosyncrasies. Design & lead dress also present issues.

A lot of the very early sets had all had paper type filter caps. When electrolytics came a long they did not pass RF very well and needed a non polarised cap to decouple that RF. Many are unaware of that when using Electrolytics to replace the paper ones. Several in USA are recommending Film type caps instead if electrolyltic.

One thing to beware of is RF in the area of the 2nd detector. With modern caps lead dress & radiation has to be considered in that area & shielding becomes paramount. It is also critical that standing waves not be allowed to develop.That is more common in transcievers. This was well recognised early.

The autodynes with their "Plate detectors" often had RF chokes between them and the next audio stage. There was sound logic here. Tubes often are not fussy about what they amplify. If you let RF into the audio, then you may find yourself with several watts of RF flying around causing havoc & seriously degrading the AF signal. I have had cause to put an RF choke on a tube to stop that.

On standing waves: If precautions like shielding are not taken, a long wire run can be an antenna, or a radiator. In the odd set you will discover something like two tubes on the same rail with an individual decoupling cap on their screens. Do not change that, as in manufacture they realised they had a standing wave issue and that fixes it.

Hi Marc, all of your last makes sense.
Standing waves, feedback loops, de-coupling, earth loops.......................

That is why I elected to put each stage in separate alloy boxes and sprinkle bypass caps of both poly and ceramic from active pin lugs to a earth loop around the valve socket. I was very concious of feedback loops forming along the heater wires, HT leads and bias feed.
In short anything that went from valve to valve. Hence multiple ceramic disc caps to ground.

Getting closer to your experience, I had an "oscillation" of microsecond duration forming on the screen grid of the first valve even though it was clearly bypassed to the ground ring with 2 caps! There was something going on there similar to what happens in a deliberate pentode Puckle type oscillator circuit as used in sweep generators. The valve would go "mad" at about x15 gain and produce a saw tooth burst. Bloody hell!
I guess there is not much you can do about the length of valve structure from the base pin up into and inside the valve.
Oddly I shut the thing up by putting another poly bypass cap from the screen pin to another ground point. Three caps going to different ground points! I know there are sorts of signals flowing inside and around the valve itself that have little to do with logic!! Just another thing to quickly "fix" and move on to the next problem!

Thanks for thoughts.

Fred.

Some times I could kill Edge: So this is from Firefox.

The worst tube I have had anything to do with is 6U7 (aka 6D6 aka #58 different heaters & bases) it unshielded is a horror. I have mentioned frequently, that my major issue is "White Noise" from the miles of 3 phase transmission lines acting as antenna's & bringing in RF. This has resulted in filters with spike MOV's being made. Some are modified inductive types.

Do not overlook ferrite beads. With heaters being AC do try them. There was merit also feeding them with a twisted pair.You will note on many quieter sets the feed to heaters is staggered. If (Octal) 2 & 7 (main heater pin out: Not all) on V1 is 2 & 7 the next will be 7 & 2.the idea behind that is to reverse the phase & cancel it.

HMV 661was constructed such that it basically had its parts in a tunnel virtually forming a Faraday cage, whist at the other end a Midwest with ineffective cage & wire carrying RF passing through the audio and audio wires passing un-shielded points if the cage. Best a one off "Golden handshake " Variant of a production run AWA R301. Instead of a separate PP amp. Its transformer had been heavied up and a single 6A6 used with the original OP tube being self biased & used as the PP driver for the splitter transformer.

Only set I have ever had with DX in three stages and AGC & a bandpass IF1. I actually reverse engineered it to get the circuit, albeit the parts were listed?.

Also the only set with no signal input, that was dead silent. Performance! amazing no expense spared making it.

Marc

Excellent suggestion about twisted pair cable. Perfect cheap filtering of stray RF which always seems to find its way in DC power supplies.

Twisted pair power supply wiring was mandatory in RF relay equipment installed in Telecom repeater stations.

We made our own on site by running two cables to the required length, fix them at one end. e.g. in a vice, and the other end would be inserted in a variable speed drill chuck. Good fun and a very effective HF, VHF and UHF decoupling capacitor built into the power cable.

Part 4 uploaded.

Hi Fred,

Regarding your concern with interconnection cabling between each RF amp:

I can see relatively large orange/red/yellow cables in each unit. Are these the cables you refer to? Are they heater and HT supply, I am assuming they are?

I see some rather large conductors here running parallel to each other. Nice capacitor plates, huh? We can bust that up pretty nicely by twisting them together. The twisted cables must remain so out of each box and to their destination (power supply?)

This breaks the relatively long run of parallel capacitor plates (that is what those cables look like to RF; a nice pF cap to hitch a ride on and escape from the box) A super highway straight to the big smoke.

Another diversion re things that go wrong from the word go: I managed to win a severely hacked STC chassis 59. Highly useful as no circuit could be found. The volume control was also missing & the hack was to put it in the grid circuit of the OP tube. That then required the re-building of the bias train, of which, being an early Autodyne the volume potentiometer was a critical part of it. Abject Fail.

I reverse engineered it & during research found what value the pot was: 6K. So I ended up using 5K WW. The circuit is now with HRSA & Kevin Chant with a warning on it as to how it was created. DX switch was also missing.

It was notable when looking for lost signal that the Oscilloscope spotted a stray V1 oscillation caused by an oscillator coil wire crossing the parallel screen & plate wires, at an angle, not 90 degrees. To kill it I ended up moving the screen wire & putting it in shielded cable.

I ended up using a series 68pF for DX (I have very strong stations). With the Plate detector in spec. I ended up increasing the bias of the IF 6D6 to neg 4.5V to stop it overloading the Plate Detector.

Marc.

Hi BBVT, those are welcome and accurate observations about the cabling mess.
By the way, I do not mind anybody commenting on any thing I build.
As a "designer" I always took note on any input from anybody, so guys put in your comments, they are not silly, there is always a chunk of gold or a reminder in what anybody says.

Back to the "cabling". Yes, there are about five wires needed to feed and water each amplifier stage,
As usual I used whatever bits of wire were handy in the scrap box so the size of wire is needlessly large!
However, those connections should emerge from the shield boxes via ceramic feedthrough connectors to by pass the RF to ground.
The inside of each shield box is glowing with RF and my crappy cables are picking the RF up and radiating that out into the under-chassis.
That is what limits the overall gain at the moment. The handshaking of RF between the stages reaches a critical point, after that the stages turn into a latched up oscillator. Only shutting the power off can unlatch it!

As I work on the signal generator half of the project I can plan a rebuild of the microvolt meter.
That will include re-mounting the shield boxes upright.
That will sit the valves vertical, allow me to include some vent holes and fit the bottom of each box with feed through pins bypassed with ceramics to lower the radiated RF.

Yep I take the point with twisted pairs as well, and that is easy to do with the heater, HT and bias supplies.
The inputs and outputs between stages will pass side ways as they do now, with one BNC input connector, lose the inter-stage BNC test points and one short exit wire with the rectified DC output level.

Like I say, any comment welcome.

Fred.

It is amazing where RF creaps in. As I have noted manytimes the HV transmission lines, due to the number of transmitters around, have RF riding on them. That has to be removed at the PSU stage.

It is to be noted that some of that RF is capable of getting into the heaters. That risks a transfer to the cathodes. As noted berfore; Ferrite beads & coils were often used on motor driven inverters & vibrator power supplies to decouple RF and brush noise.

The coupling to Plate detectors via RFchokes, is common and I have done simiar to some tubes to dampen their enthusiasim, for amplifying RF to the detriment of audio quality.

Marc

It pays to remember that there is no such thing as a perfect capacitor. Wound polyester caps have significant inductance in series with them. This varies with the value of the cap.

Even surface mount ceramic capacitors have inductance.

A design I'm working on at the moment has 5 processors in one chip, about 20mm across. It has 14 separate power supply rails, supplied from a special purpose multi regulator chip called a PMIC. Parts and cores of the multi-processor chip get powered down when they are not being used to save power. There is a 3GByte RAM chip and a 48GB flash drive. It has many clocks that run in the 1GHz region.

All the power supply balls on the chip are bypassed with at least 3 and usually more ceramic caps, ranging in value from 22μF to 0.1μF. Most are 0201 size, which means they are 2mm long by 1mm wide. But even this is not enough. the inductance is still too high. So the trick is to use the inter-plane capacitance of the PCB (it has 10 layers) to form a capacitor. The inter-plane caps catch the first edge of the current pulses and the other caps fill in, fractions of nanoseconds later.

All this on a PCB about the size of a credit card.
Your mobile phone uses similar technology.
I relate this to observe just how much the technology has advanced!

Well, here is a bit more work done on the micro volt RF meter.

I stood the valves up, removed a lot of excess grid wiring, heavy wire under chassis connectors and generally tidied up the build.
The meter works better by a decimal poin tnow and can measure micro voltsRF levels with repeatability.
However I have run headlong into the problem of background RF, RF from local radio transmitters, and general switching crap flooding the ether and the workshop.

Tesla or Marconi never had this problem!

While there is "unauthorised" RF blasting away, leaking into equipment it is very hard to get down to those micro volt levels I would like to explore.
I did do a whole year of sub and standard testing at tafe/uni many years ago and I wish I had the caged lab that work was carried out in the backyard!
Anyhow the meter performs down to its own generated hash floor level, but at best in the 10's of microvolts, nothing less.
I did also wrap the meter chassis into a metal box but this did not make any substantial gain so I guess the meter stands as it is, an interesting exercise but not much use.
I'll send a PDF to Brad covering this work and it can go with the rest of the PDF's at the start of the thread as "AM sensitivity part 5"
Fred.