How to get a Univox U-45B so quiet you can't even tell it's on: A Tutorial

[SoulFetish]

... I mean, so quiet, I had to play it again just to make sure it was working. Here's the story:

If any of this sounds familiar, it's because when I searched for information about this model, there are a few specific problems that owners and technicians experience which seem to come up over and over again relating to "noise". In my example, there were three sources for the noise problems, so I'm probably going to have to break it up into a few different posts.

First: On the ones I've seen, the idle hum is the biggest point of concern for most customers. Often, they request a "cap job' to replace the original electrolytics that were stock to the amp. This isn't unreasonable, and is often the first thing a tech might look at with an amp of this vintage. However, in these models, you can often replace all of the electrolytic caps outright, and even increase the capacitance thinking it's a lack of filtering only to have no effect on the audible hum level. This is how it was with the both of 2 models that recently came in for service (with the second coming in about 2 weeks after the first).
We noticed fairly quickly when testing the amp, that the hum start the moment the switch was turned "on", even before the tubes heated up. So, we pulled all the tubes, powered it on, and the hum was still present! This left the most likely source being the leakage flux coupling through to the output transformer, which was exactly the case. The only way to fix this is to remount the output transformer. Realistically, there are two ways to do this; First, mounting it on the bottom/outside of the chassis. We weren't sure if it would fit using the existing mounting holes with the speaker frame. The other way, which we did, was to rotate the OT 90˙. This can be done without having to do a bunch of rewiring and drilling, if one is very careful. We were able to use one of the existing mounting holes and an L-bracket to position the OT and completely null the radiated PT flux.
It's easier to show a couple of illustrations of how this worked. The first is an overview of the stock manufacturers layout, and the second is the position in which we were able to achieve the lowest noise:



If you decide to remount it internally, I would advise that you follow these steps to make the job very easy. I happened to have some brass L-brackets which fit nicely, but you may have to do a little searching or small amount of fabricating.



This solves the first problem.
I encountered 2 more which I'll write about, and how to fix them shortly...

[Jazz P Bass]

Nice!

[SoulFetish]

Part II.

Ground loops...

Long story short is this amp has them, and other strange grounding issues as well. Before I get into fixing the grounding scheme, I should touch on something which I can only assume was an oversight in the design of the input. In the stock design, all three input jacks are the switching/shorting type with a ground bus wire connecting all the grounded and switching terminals in all three non-isolated jacks. Because all unused jacks are shorted to ground, when you plug into the guitar input, the other two shorting jacks create a voltage divider from the guitar inputs 47k grid stopper to the other two 47k resistors in parallel making the low side of the divider around 23k! This not only causes a heavy loss of input voltage, but puts a heavy load on passive guitar pickups as well. See stock schematic here:



After this mess, the input is capacitively coupled to gridleak biased input stage. I modified the input as shown in the schematic I redrew below to remove the unused inputs, changing the grid stopper to 10k, and cathode biasing the input stage with an unbypassed 1k5 resistor. Ultimately, even though leaving the input unbypassed is a little unconventional, I found that the amp had better overall tone and more pleasant overdrive when pushed harder. I think there's room for experimentation here for those with the time to spend on it.
Here is the modified drawing. The notable changes are to the input as described above, increased filter capacitance, and 100k gridstopper on the cathodyne PI input to moderate the overdrive characteristics.


The following grounding layout and instructions were the result of getting pissed off at the prospect of not solving the noise issues and being put onto another project, so I stayed until 4am on this til I finally had all the issues resolved. So, I'm just going to post the layout diagram, and try and answer any questions if there are any. Let me know if anything looks off, or if you catch an obvious mistake so I can fix it. I apologize if there are any spelling errors, I tried to explain any changes and get it as close to scale as I had time for.

[SoulFetish]

Actually, I caught an error. When I copied and pasted some of the schematic symbols I forgot to change the EL84 label to 6BM8 on the pentodes. Ill change em later and upload a new image

[SoulFetish]

updated schematic with corrected tube labeling:


I'll follow up on this and wrap up my findings shortly.

[trobbins]

I guess one option could be to connect the heater CT to the output stage cathode, as a simple form of DC elevation. That may possibly also benefit a 6X4 fault where cathode shorts to heater, by reducing fault current level.

[35L6]

No grid return on the input ?

[SoulFetish]

No grid return on the input ?

I'm sorry, I don't understand what you're asking me?

Oh, I noticed on the schematic layout drawing, I didn't add the 1M grid leak resistor on the input jack. I did add this to the repaired amp, but I'll make those changes now and correct the files. Is that what you were referring to?

So I'll upload the revisions here, but I would ask that a moderator would give me the ability to edit the original posts to keep all files organized for version control. I don't want someone to inadvertently use files that contain errors.

[Enzo]

You Seem to be relying upon the pickup to be the grid return of the input stage.

[SoulFetish]

I guess one option could be to connect the heater CT to the output stage cathode, as a simple form of DC elevation. That may possibly also benefit a 6X4 fault where cathode shorts to heater, by reducing fault current level.

I certainly like the reasoning behind the suggestion. I can't speak to how it might affect the noise performance. If one were connect the Heater CT to the output tube cathodes, I would certainly increase the bypass capacitance of the cathode resistor from 50µF to 100µF or higher. It's important to try and keep that voltage reference as stable as possible or it's going to reek havoc on the stability of the amp when it gets pushed hard, and one of the output tubes goes into cutoff. This might be even more critical in this particular amp because 2 of the preamp stage triodes share an envelop with the output tube pentodes. An alternative, you could string together a series of MV50152 LEDs from Cathodes to ground to achieve your desired bias voltage. That LED won't drift more than 4mV from around 10mA to 100mA(max rated continuous I_f). (It could look something like this):




I will say I actually used a hum balance pot after disconnecting it from the HV center tap, but found grounding the heater CT as described above to be quieter.
Another way of protecting against damage cause by a heater short in the rectifier might be to install in-line fuses on each of the legs of the HV secondaries.
There are probably a number of changes that this model might benefit from. The procedures for listed above, however, were limited to examples which were tested and confirmed on working amps to provide the quietest operation.

[SoulFetish]

You Seem to be relying upon the pickup to be the grid return of the input stage.

Right, it was a mistake in the drawings, not the repairs. I've revised them to include a 1M grid leak.

[trobbins]

50uF cathode bypass is capable of keeping the heater elevation pretty quiet for mains and above frequencies, given that the elevation's main aim is to suppress 12ax7 heater-cathode hum if the 12AX7 had a low heater-cathode resistance - but yes if the elevation has any signal on it then the higher signal frequencies may become noticeable due to leakage across the heater-cathode capacitance, and unexpected wiring loops.

I'm all for adding protection - adding 1N4007's to each valve diode anode would be my suggested first and simplest effort. The list goes on, but in the end it is a risk/effort tradeoff.

[Helmholtz]

An alternative, you could string together a series of MV50152 LEDs from Cathodes to ground to achieve your desired bias voltage. That LED won't drift more than 4mV from around 10mA to 100mA.

The LED datasheet shows a Vf increase of 0.3V-0.4V between 10mA and 100mA. Multiply this by the number of LEDs.
https://eu.mouser.com/datasheet/2/14...31-1165942.pdf

I don't like using diodes to replace the cathode resistor in power stages.
It turns self-adjusting auto-bias into (non-adjustable) fixed bias.
While auto-bias stabilizes the idle cathode current by DC NFB, diodes don't and a different pair of tubes may run too hot or too cold.

When the tubes leave the class A region, average cathode current and bias voltage across the resistor increase and counteract tube overdissipation - but hardly so with diodes.
Also LEDS have considerable Vf variation (e.g. +/- 0.2V) and Vf drops with increasing temperature.

[SoulFetish]

The LED datasheet shows a Vf increase of 0.3V-0.4V between 10mA and 100mA. Multiply this by the number of LEDs.
https://eu.mouser.com/datasheet/2/14...31-1165942.pdf

Yes, that’s right. I said 3-4mV, that was a mistake. I should have written 300-400mV. But if you look at the datasheet, they obviously going to operate somewhere around half that range. So, 0.15V-0.2V from class A to B operation? Multiply that by the number of diodes. Thats a fairly linear device.

When the tubes leave the class A region, average cathode current and bias voltage across the resistor increase and counteract tube overdissipation - but hardly so with diodes.

Exactly what you don’t want if you plan to connect your heater supply as a voltage reference. If you connect your heater CT there and the cathode voltage modulates much, you’re going to wish you hadn’t

[Helmholtz]

Yes, that’s right. I said 3-4mV, that was a mistake. I should have written 300-400mV. But if you look at the datasheet, they obviously going to operate somewhere around half that range. So, 0.15V-0.2V from class A to B operation? Multiply that by the number of diodes. Thats a fairly linear device.

Diodes (including LEDs) are as non-linear as it gets. Nothing is more linear than a resistor (meaning that dV/dI = constant, ensuring zero distortion).

As explained above, it is desirable that the cathode voltage increases in class (A)B operation to avoid overdissipation, especially if idle dissipation is above 70%. Diodes/LEDs might not allow for sufficient increase of bias voltage.
Otherwise you will need lower idle currents, i.e. something like 70% plate dissipation or lower at idle as typical with fixed bias.

Exactly what you don’t want if you plan to connect your heater supply as a voltage reference. If you connect your heater CT there and the cathode voltage modulates much, you’re going to wish you hadn’t.

Have you seen problems with heater elevation using the cathode voltage with a bypassed cathode resistor?