Sir John . . . . .

My Grunow model 4C still hums with the volume all the way down after replacing the filters

That sort of clears up and elimates the front of the set and suggests only the AF output stage or its problem being power supply ripple related.

BUT . . . your further researching that one portion ? of the power supply was only exhibiting but 700 mv of effective AC ripple.

I usually don't hear hum creeping in and onsetting until up in the 3-4VAC thresholds on table models with their 4-5 in speakers, but, that level would be more emminently noticable on 8-10-12 in speakers.

I'm placing a thumbnail mark-up of the suspect problem area here, and will be giving some related comments on it:

Initially in sleuthing, why don't you try the taking of a 47/50 ufd electrolytic of at least 50 vDC rating and have its positive lead going to chassis ground near the AF output stage, so that you will then be able to able to reach the 1st'us gridamus ( #4) of the 42 AF output tube with the caps negative lead and then touch it to that grid which should cause an initial "pop" and then that should have everything coming in from the speed of light (c) all the way down to ~.01 Hertz being shunted off to ground.

Which then NOW means, that IF your hummmmmmmm is still there it should be the 42, its adjunct circuitry or being derived from the sets power supply, its not originating externally and coming into the 1st grid, nor is it being amplified within the 42.
If you have not tried another 42 to additionally to check for the possibility of any fil/cathode leakage within that tube, one can improvise in testing the initial tube by connecting the two supply leads of the filament pins of the 42 to their respective supply wiring thru 2 clip ended jumper clips and then warm up the set until hummmmmmmm onset, and then pull the 2 clip leads that are clipped to the tube pins and run on the 5 or so seconds of then floating filaments with their latent heating, in order to aurally evaluate to see if the hummmmm has then disappeared.

No joy yet ?

Now let's be consulting the mark-up thumb schema and review the manner of derivation of the 1st grid bias voltage for the 42:

The HV centertap of the high voltage winding is needing to get grounded, but in this sets design it is taking a path thru the field coil of the speaker to chassis ground to both use that inductance for power supply filtering and the magnetic energizing of the speaker core PLUS creating an additional supply voltage . . . . it being a negative supply to be exact.

Trailing the YELLOW buss from the HV C.T. we see it initially passing down to, and then thru the speaker field coil to ground.

That speaker field coils 2000 ohms of resistance is holding the HV C.T. up from ground, so that a potential negative voltage of up to ~ negative 80 volts could be at the C.T. with only minimal power supply loading being placed upon the set.

Look at the item [ 21 ] filter capacitor and we see that it is providing the initial filtering of that negative supply.
In this set they have opted for the placing of a tap on the field coil winding in order to voltage-divide down by a fourth to get the the actual amount of voltage needed for the biasing of the 42's 1st grid, which is then flowing down the FUCHSIA buss to an initial isolation resistor [ 8 ] and a bit of high frequency filtering and bypassing by .25 ufd capacitor [ 15 ] and then feeds the [ 7 (right) ] 42's grid resistor which then feeds bias voltage on into the 42.

If you found no K-fil leakage in the 42 and the sole [ 20 ] main B+ filter is up to par /adequate then lets look at the further incorporation of its hum bucking technique, because.

I noticed that the leads to the speaker have been cut and spliced back together.

If that is the case and if they were reversed, that is a possibility of your problem.

Look at the small red Insets, giving the phase relationships of a single sine wavelet of your hum as its origin at the 1st grid of the 42 and then the plate and output transformer circuitry all the way to the speaker.

(In your case, if the hummmmmm was STILL present with the 1st grid capacitively swamped to ground, then we can assume that the humnnnnnn is just derived from the field coil voltage having ripple in it.)

The hum bucking coil is dependent upon its amount of turns and physical / mechanical proximity to the field coil proper in establishing its amount of pick up.

It is wired in series with the speakers voice coil and then one consults the larger BLUE inset chart to see the potential problems if there is a phasing problem between the hum buscking coils sample and the hummmmmm being presented to the voice coil from the output transformer.

If the top illustrations conditions are met the humm is almost cancelled or greatly minumized .

The bottom conditon just worsens the situation, as you then have two "hummmms" being summed.

Now with your one time broken connection and tape repair of the speaker wiring connection . . . assumably to be my two inserted RED x's.

It would be difficult to detect any minor definitive evaluation with the shutting down of the set, in order to reverse the connections each listening session.

To make an almost immediate evaluation , I have a DPST toggle switch wired for polarity reversing as shown below with 4 leads and clips connected, so that flipping the switch will instantly evaluate the two reversed connections . . . thereby, having no waiting disruptions.

All of this info may already be in your knowledge base . . . but there just might additonally be the uninitiated person that is silently lurking " up in the peanut gallery " and also benefitting from its info, with their read.

Pls . . .re-post us on your findings ! . . . . .

73's de Edd