Darkvoice and Crack Mods

I am ressurecting this threat thanks to the generosity of @roshambo123 giveaway and also allowing me to purchase a couple of upgraded tubes for this DV. You see, I got rid of my DV 336SE long ago in other audio pursuits. Now I have a better job and this is not my main amp and can afford to be out of commission a few days for modding.

I am also just now realizing that looking at the circuits of the Crack and the DV; straight out of the box, the DV is starting from somewhat higher place than the Crack. Mostly:

• Higher overall power capacitance
• Film caps already in the audio path for AC coupling

The Crack only has three 220µf caps on the PSU B+ (660uf total), while the DV has three 800µf on the PSU B+ (2,400µf), plus the DV added two more 200µf caps on the B+ line -one just for the power tube, one for the preamp tube. So 3,200µf total in the DV vs 660uf in the Crack - that is like 4.8x the overall capacitance.

Then, the Crack as electrolytics in the audio path on the final output for AC coupling and also setting the final output impedance. The DV uses film caps for this out of the box. Yes, you can mod the Crack for this, and probably more room to roll more exotic caps as well. However, I am not in any hurry as these are already propylene caps in the DV.

The most damming things and the greatest weaknesses in the DV is

• the transformer dumping 7V on the heaters causing hum for many tubes
• the crap linear taper volume pot with bad channel imbalance areas down low, and with it being linear, it means you are almost always in this area.
• cathode bias resistors on the preamp tube

The reason for the 7V on the heaters is because the DV 336SE has a 110V power transformer. It is no secret. Go look it up on the Drop page or elsewhere. It is clearly listed. When I put the DV on my variac and dropped it down to 110V on the output, I saw 6.3V on the heaters almost exactly, and no hum on the tube I was using.

So yes, I plan on:

• putting LED's on the cathode on the preamp tube with 2V forward voltage LED's
• get the 7V down to 6.3V on the heaters by using a voltage divider network of two resistors (100ohm and 1K)
• change the volume pot to Alps RK27 50K.

One side of the heaters is tied to ground, but there is another trick you can do with the heaters because there is no center tap on the heater is to create a fake one with a couple of resistors and tie it to the cathode bias voltage of one side...I did this in a guitar amp with much success. I will need to look up this circuit again and figure out how to implement this in the DV. If I do this, then I would need I think a 4k resistor for the power LED since this has two 2K resistors now on both legs.

 

So I think I fixed the hum issue without changing the bias or whatever. Even without fixing the high voltage.

I realized they are grounding one side of the 6.3V heater rails. It appears there is no center tap for ground. So I disconnected the ground on that one leg that is close to the front LED, then I did the virtual ground with two 220ohm resistors.



I have a new production EH tube that will not work in a stock DV. Hum city. Doing this works. Quiet as can be. And I am still measuring 7V on the heaters.

Now I could also do this and instead of float the heater ground on one of the bias of the cathodes.



But not sure that is necessary in this small simple amp. I’ve done this in guitar amps that have much more going on and more sensitive to hum. The ground seems to work great.

Now I racked my brain on how to combat the higher unregulated voltage with a 110V input transformer without having to change out the transformer. I could use a couple of diodes to reduce the voltage on the heaters by 0.7V, which makes it right at 6.3V. But that does not address the higher plate voltages and higher cathode voltages for proper operation point.

So I figured why not drop the mains by an inline resistor or two to drop 10V so the transformer sees the proper 110V. Well, there is a certain balance between power draw and voltage drop to figure what exact value you need. With some experimenting, it seems I need 33ohm resistor. It will need to be 5W resistor or more, and I better go even higher wattage than that really. So more on that later when I get the right resistors and get this completely figured out.

In the meantime, also enjoy the amber cathode bias LED’s.

 

UPDATE:

Yes, I did this:



Every Darkvoice owner should do this. It is really easy to implement as the ground is right there and you can do it just like the diagram here is. Just get two 220Ω 1/4 watt resistors and wire from the filament to the ground buds wire.



And then disconnect the wire ground that is close to the LED. You don't need two grounds, but one leg tied to ground is really what is causing the hum in some tubes, along with the higher heater voltage (more on that below)

The other thing I did was this:



The DV has two 2K resistors on either side of the LED. They will need to go in front of that. You can solder it right on the same point as where the LED soldered on that tiny LED board. You just have to experiment with how to orient the 1N4148 diode. If you have it the wrong way it will just make the LED go out. If you have it in front of the resistors then it will blow out the diode. I really was not going to do this, but I had a few 1N4148's laying around from a PSU build so I gave it go. It does not really help out audio. According to Valve Wizard, "This diverts reverse current around the LED and keeps the reverse voltage across it to one diode drop (and it also reduces DC magnetization in the transformer)."



The next thing was the LED bias. You can go for 1.7V or 1.8V forward voltage LED's. Just connect the anode to the tube socket and cathode to the ground replacing the 1K cathode bias resistor.

The next fix was putting an RK27 Alps 50K log pot. The one installed is actually like 82.5K or something and it is linear. Forget having it play quietly as a desktop amp. Now with the RK2K, it is very nice. I also replaced the knob over the plain stock one. I will say this swap out was not easy. You either need to cut off the anti-turn knob or be crazy with the drill like I was. I really should have just cut it off. It also barely clears the octal socket behind it and really rests on it just barely. And figuring out the grounds was also fun, so be prepared to change out the wires that go to the screen inputs from the pot. It is a MUCH needed fix for this amp though unless you just like it loud.



THE REAL FIX:

Alright, I mentioned before this transformer is supposed to see an input of 110V and not 120V (123V at my house most days). This means it is dropping close to 7V on the heaters. This is about the limit of the heaters because most are rated +/- 10%. And 10% high of 6.3V is 6.93V. ((6.3 * .1) + 6.3)) So, how to fix this?

I thought of using a couple of diodes to eat up 0.7V like this:



But I also figured the operating point is also kind of high. Plate voltages high, etc. And I checked and it was. So one way to fix is just have the right ratio transformer for the correct input voltage. But what if I took away 10V or so with a resistor or two on the input. I started doing calculations of load and what voltage dropping resistor I would need for the load of this amp and how many watts I would be eating up. I won't bore you with all my experiments, but I settled on needing about 30Ω resistor with eating up about 7 or 8 watts. I also realized my input power is higher than most and I need a 13V drop and most just need a 10V drop (if you have normal 120V power). So I came up with the following calculation (for this amp only):

Voltage drop needed * 2.5 = Resistor value needed
So for me: 13V drop * 2.5 = 32.5Ω
10V drop (for 120V input) * 2.5 = 25Ω
​

I also realized that the wattage of the resistors is normally dependent on attached to a heat sink. At least the wirewound ones above 20W or so. These also needed to go IN the amp below the other power resistors between the caps that are already creating lots of heat - they can put out about 80C. More experimentation and actually settled on two 60Ω 50W resistors in parallel, giving me 30Ω of total resistance. And now my heater voltage is down to 6.27V and the resistors are only kicking about 55C. Much lower than the 80C daddies above them (those are only 10W). So if you have normal 120V power, then two 50Ω 50W resistors will do just fine.

Just disconnect the power going to the power switch, route it through the resistors, and route it back to the switch. I secured my resistors with some thermal glue. Be sure to heat shrink or cover everything...that IS a live wire.



All i can say is that it is finally nice to hear this amp for what it is supposed to be. It is dynamic as hell and I actually think it is cleaner now that operation is closer to the original design.

NOTE: This DID increase the temp of of the outside of the case by a little bit. After on a few hours at room temperature it finally settled around 125F. Which is not all and all terrible, just a bit warm. Still lower than a first gen. Asgard.

 

Glad this project was so successful. I hope others with the amp are able to gain knowledge.

@Azimuth In a couple sentences could you summarize how the virtual ground with 220 ohm resistors is superior to the 'Fitz Mod' ?

 

You can read about heater supplies in more detail here:

But to summarize:

"Hum and buzz is caused by leakage current between the heater and cathode, and by the electromagnetic field around the heater and associated wiring. The magnetic field modulates the electron stream inside the valve, leading to hum. The electric field couples into the audio circuit (especially the grid and anode) via stray capacitance. You might think that the stray capacitance would be too small to couple low-frequency hum, and you'd be right, except that the mains supply -and therefore the AC heater voltage- is not a very clean sinewave. The peaks are usually clipped off by rectifier action, and who knows what industrial motors pollute the supply with hash. These ugly corners and spikes contain high-frequency energy which couples quite easily through even tiny capacitances. This leads to more of a buzz sound than a low hum."​

So there are few ways to deal with it. The most common is to have a transformer center tap on the 6.3V section:

(the resistor here is just acting like a fuse in case of tube failure that will protect the transformer. You can use anything between 10-100Ω flameproof resistor)

Although the DV transformer does not have a center tap for the 6.3V heaters, and the makers of the DV just grounded one leg of the 6.3V rails. It is true that the heater rails need to be connected to ground, but I have never seen any modern amp just ground one of the rails and not have a center tap or virtual ground. The virtual ground is the most common you will run across and what I did here with the 220Ω resistors.


You can do even more tricks that you can read on that Valve Wizard page. Lifting the virtual ground above the heaters using the same voltage as the bias voltage, using a hum balancer, just doing away with A/C heaters and just rectifying down to DC, or just DC the first input tube, and then covers hum wiring, which the DV does go through the center of the socket, although they are not twisted, but I digress.

Then, you have to understand what the Fitz mod was doing (with the 7V heaters and one leg grounded). But basically only really increasing the gain with a bypass capacitor on the cathode bias resistor. That increases current on the cathode to be above the hum. So it never really addresses the issue of proper grounding scheme of the heater rails that we have realized for years. I am just surprised that this amp has been out this long and I am the first person to actually do this modification and to fix it properly.

NOTE: the Bottlehead Crack has a transformer center tap on the 6.3V heaters.

So thank you again @roshambo123!