Osram Music Magnet 4

[Ed_Dinning]

Hi Sing, an old valve can be run at a lower bias to get up near max Ia, but distortion levels will be high. It may be better to us a slightly lower bias and accept lower sound output and acceptable distortion.

Ed

[Herald1360]

Adding a 1k shunt resistor will drop Ia for a triode unless the supply voltage is increased to maintain the same actual voltage on the anode as without the shunt. That's why I said measure the actual anode voltage, rather than assume it is the same as it was without the shunt if you do it that way, when you compare with datasheet values.

I didn't twig that you were trying the PM2A output valve in a regen circuit- I assumed you were checking its audio output level in a straightforward single ended class A audio amplifier circuit- as it would be used in a multistage set using a PM2A power stage.

[regenfreak]

Ed yes the PM2A indeed gets fair amount distortion running near max Ia.


Chris, I have built many Homebrew regens before. My current audio output valve "test bed" is a Twinplex regen using a TV/radar 6SN7GTB (that thing is pumping out lots of heat):

https://faculty.frostburg.edu/phys/l.../twinplex.html

It is very sensitive and produces loud output. It has rather poor selectivity that I have built a 0.9 meter diameter hexagonal loop antenna to improve its selectivity. I live in a flat in Southfields, London, the home of the great Captain Stanley Mullard; the MW reception here sucks though. The local Bollywood channel overpowers other stations

I wire the same way like the Music Magnet 4 circuit after the detector stage. The Osram MM4 detector stage is very similar to many one valve regen sets. See the attachment as example.

I am still not sure why the original Gecophone transformer has the high impedance side (dc resistance 1.87K) marked as primary and low impedance side as secondary. It works like a step down transformer which does not make sense to me. I can take a photo if you don't believe me. Someone asked exactly the same question before, apparently the "wrong answer" by Alan is supposed to be "right one"
Here was the thread:

https://www.vintage-radio.net/forum/...ad.php?t=67429

[regenfreak]

Originally Posted by ms660

It doesn't need one, but a regen detector will.

Lawrence, I thought the exactly the same but I didn't believe it until I tried it myself.

But Ed has said the DC resistance of the interstage transformer is quite low, it does not allow the build-up of grid charge and 1M resistor is needed

I need to go away to read and think about this.

[Ed_Dinning]

Hi Gents, slightly wrong. The low resistance of the IV trans allows dissipation of charge on the grid. In RC coupling the "grid leak" must be fitted.
As a belt and braces test it eliminates grid charging and will have little effect on sound levels in the op stage, purely a method for eliminatingfaults.

Ed

[Herald1360]

The circuit posted is of an interesting low voltage HT set- the 49 was a bit unusual and suited to this role. The tapped HT feed to the 49 would vary the gain which would work with the fixed level of positive feedback (reaction or regeneration) to do this. The output stage is running at zero bias and isn't representative of a typical PM2A application which would have the earthy end of the transformer winding to the grid connected to a -ve voltage from a grid bias battery.

An audio transformer with a stepdown would possibly be used to drive a zero bias following valve where grid current would otherwise cause distortion or it may have been intended as an output transformer to match the optimum load impedance of an output valve to a lower impedance loudspeaker.

[regenfreak]

Originally Posted by Ed_Dinning

Hi Gents, slightly wrong. The low resistance of the IV trans allows dissipation of charge on the grid. In RC coupling the "grid leak" must be fitted.
As a belt and braces test it eliminates grid charging and will have little effect on sound levels in the op stage, purely a method for eliminating faults.

Hi Ed
I only understand you now what you meant by "belt and braces" measure. I must clear my head and repeat the test on the PM2A all over again. When i applied any negative grid bias directly without 1M resistor, the sound was completely cut off but the anode current does not drop . I thought I was doing something stupid mistake but the set-up was simple enough.

Originally Posted by Herald1360

An audio transformer with a stepdown would possibly be used to drive a zero bias following valve where grid current would otherwise cause distortion or it may have been intended as an output transformer to match the optimum load impedance of an output valve to a lower impedance loudspeaker.

Put me right if I am wrong. If the PM2A can easily self-destruct and burn out the interstage transformer with too low grid bias voltage, say zero volt GB at 130V anode voltage, then there must be large grid current flowing to the anode under this condition. The hypothesis here is that : the PM2A has low input impedance which may explain the counter-intuitive, low impedance winding for the secondary of the MM4 interstage transformer.

Like what you wrote, the 3 stage TRF has high amplification factor already, so it is no big no deal to have the voltage dropped by a factor of 3, for it to match a low impedance loud speaker. Probably the wireless experimenters in the 1928-1930 wouldn't want to folk out money to buy another audio matching transformer for the horn speaker. In 1928, the MM4 cost £10; inn today's money equivalent to £630 .

[Herald1360]

That sounds a bit confused......

With zero grid bias, a higher than normal anode current will flow which may damage the PM2A if the anode voltage is high enough for the current to exceed the max anode current rating for the valve. This current flows from anode to cathode and doesn't relate to grid current, only to the anode-cathode voltage and the grid-cathode voltage (bias voltage).

With zero grid bias, current flows from grid to cathode only when the grid is positive wrt the cathode as will occur on positive half cycles if the grid is driven by an ac (audio) signal. To the valve driving the zero biassed stage, this means that for the negative half cycle the zero biassed grid looks like a high impedance since the grid-cathode circuit looks like a reverse biassed diode but for the positive half cycle this diode is forward biassed so the grid passes current and the impedance seen by the driving stage is much lower.

Normally this state of affairs is avoided by never letting the peak positive drive voltage exceed the negative grid bias voltage but there are some output circuits where this is allowed to happen. E.g. it can be a way to screw as much audio power as possible within the valves' ratings out of a push pull pair. Where this is done, the stage driving the output valve has to have a low enough output impedance that the varying impedance presented by the output valve grid doesn't distort the driving signal by clipping its positive peaks.

As to why the MM audio interstage transformer might be a stepdown type, I've no idea. It seems to be completely counter intuitive- if there's so much signal from the detector stage that you need to reduce the drive to the output stage, why use an expensive transformer at all when a couple of resistors and a capacitor would do just as well?

A transformer to match an output valve to a low impedance speaker will be in the anode circuit of the output valve and nothing to do with the input side of the valve. I only posited that idea as a possible reason for a stepdown audio transformer per se to exist.

[regenfreak]

Many thanks for the explanation. I understand most of what are you saying. I need to go back to read some old books about valve radio to brush up my knowledge and understanding of the amplifier theory.

I enjoy browsing through old battery radio circuits in British and American vintage radio magazines 1920-30; I am always fascinated by the engineering, art and design in that era:

https://www.americanradiohistory.com/#Else

Sometimes I have seen strange circuits that I thought that's interesting..and counter-intuitive.

[regenfreak]

Following the helpful advice of Ed, I can get the grid bias working now with a 4M pot and a few D cells in series. I think my mistake was that I applied to grid bias voltage directly to the grid of the valve instead of applying through the secondary of the interstage transformer. It turns out the valve has a quite good emission not far off from the datasheet graph.

I actually have two multimeters to measure the Ia and GB voltage PM2A emission:

At Va of 92V:

Ia = 7mA GB -1V
Ia= 5.2mA GB -2V
Ia = 3.7mA GB -3V
Ia = 2.2mA GB -4V
Ia = 2.0mA GB -4.85 speaker sound distortion
Ia = 9.12mA GB 0V speaker produces buzzing and hissing sound. The circuit goes into self-oscillation.

Speaker goes into buzzing. The amplifier stage goes into self oscillation at 0V

It was a great learning exercise.

[PJL]

The valve must have a dc connection to earth to operate correctly. For the output valve the connection is via the battery and transformer secondary. The output valve must be operated with -very grid bias or the sound will be distorted.

[Herald1360]

Originally Posted by PJL

The valve must have a dc connection to earth to operate correctly. For the output valve the connection is via the battery and transformer secondary. The output valve must be operated with -very grid bias or the sound will be distorted.

Is this referring to the output valve grid and the bias battery?

Looking at the figures given for the actual valve, I'd say it's a bit tired, but with the 92V Va I would say a bias setting of about -3V would be the best you could do with a bias battery. Optimum would be around -3.5V but that would need a fully adjustable source.

[regenfreak]

You are right about the optimum at -3.5V. At this GB, it has no distortion providing reasonable sound volume. There is no big difference in perceived sound volume between 7mA at -1V and 3.7mA at -3V. The self-oscillation phenomenon at zero grid bias is very interesting and it got me head scratching for a while.

I connected the set to power all RF stages last night with a Cyclone 2V 5AH, HT +120V, +90V screen grid for the two PM12A and PM1HF. It draws 60mA HT+ even the filaments were disconnected and I knew right away the audio pass-by 2.7 microFarad cap is leaky. I applied HT across directly across the cap and the leak current is measured to be 60mA. So there are lots of little hurdles and niggles ahead.

In the end, if I cannot get the whole set working,I won't be a big deal and I will use all valves to make regen radio sets and it is still a great learning experience.

If the whole set was working from the start, I would not investigate and seek answers, knowledge and understanding of the circuit and operation of valves. Then it wouldn't be much fun.

[regenfreak]

The differential throttle cap has low resistance of 6 ohms and is leaking massive HT current too. The SG RF capacitor is also leaking small current. The only way to fix them is to remove the old leaky caps and the set will lose its old charm.

[Herald1360]

Depending on the capacitor construction, it may be possible to dig out the original innards and hide a modern component inside the case. This retains the appearance and restores the function!

I'm intrigued by what you call the differential throttle capacitor. I'm guessing that this is the regeneration (reaction) adjustment capacitor? A picture of the part may help with suggestions for fixing it- many old style variable capacitors can be dismantled and rebuilt relatively easily.

[Ed_Dinning]

Hi the differential cap is normally a solid dielectric type that can be dismantled and cleaned as it may have an internal short.
Usual practice with the paper caps is to replace with modern types to get it going and prove the unit, then re-stuff. From memory the base of the cans can be unsoldered.

Probably best to bring up HT voltage slowly, as 60mA can damage some windings depending on the location of the faulty part. It would also flatten a conventional HT battery pretty quickly.

Ed

[regenfreak]

The throttle cap is the dielectrics cap on the right (see photo and schematic) . It is not be leaky after. It has three terminals. According to the schematic, it is a symbol of a differential capacitor. I think the two main terminals of the throttle cap is connected across the tickler coil resulting very low DC resistance of 6 ohms and the third terminal is connected to chassis ( Earth).

The RCL meter can pick up the inductance for tickler coil which is about 47 micro Henry (sounds about right). I have applied high voltage across the earth and one pole of the differential cap and both poles have zero leak current.

I disconnected the 2.7 mircoFarad leaky audio by-pass cap, something is still leaking HT current. The hunt for the culprit is still going on. I look at the schematics, it is not obvious what is the other suspect. The SG RF cap on the right of the schematic is only leaking very small current after applying the grid screen high voltage.

All the regen radio circuits I have seen, they all use a simple variable throttle capacitor connected to one end of the tickler and to Earth. I think the differential cap works in the same way.

[regenfreak]

Originally Posted by Ed_Dinning

Hi the differential cap is normally a solid dielectric type that can be dismantled and cleaned as it may have an internal short.
Usual practice with the paper caps is to replace with modern types to get it going and prove the unit, then re-stuff. From memory the base of the cans can be unsoldered.

Probably best to bring up HT voltage slowly, as 60mA can damage some windings depending on the location of the faulty part. It would also flatten a conventional HT battery pretty quickly.

Thanks Ed
I need more probing around to find the source of HT leak. I also need to check if the wiring is correct.

[Herald1360]

The differential reaction control capacitor will give a somewhat smoother and less critical adjustment at the point of oscillation since it acts as a sort of gain control (by shunting the rf at the detector anode to earth) as well as throttling the rf feedback coupling provided by the tickler coil.

One thing all these kinds of set have in common is a fair amount of ill defined rf feedback resulting from the layout of the components and on the better sets a fair amount of cut'n'try experiment would have gone into the reaction control system.

[regenfreak]

Chris
Your explanation makes sense. Many thanks!!!

I have found the culprit:

I have removed all valves and checked for shorts.

I have connected a RCL meter across the HT+ and HT-. I set the resistance measurement range to 20M. As I rotate the wave changer in 360 degrees turn, the resistance fluctuates from 20M to zero resistance ( from a poor contact high resistance short to a dead short!). So there is a short somewhere under the wave changer brass axle where there are five wires running underneath and they are physically touching the axle. I believe the rubber padding disintegrated overtime with contact friction and have bare exposure wires touching the brass axle.