dsavitsk

Sounds like a loose lam. Easiest fix might be to find what is rattling, press it into place, and put some epoxy on it. Or, I can swap you for another one.

I would -- if you don't risk it, you don't have a working hub. If you do, you might. But, I wouldn't risk it if you are not in the room in case is burns up

Looks great, Marc. I am thinking about changing the bias mechanism somewhat drastically. As has been discussed, the WE connection is not my favorite -- I think it sounds a little flat which is why I keep asking everyone about it. Additionally, I think the bias needs to be over ~2.5V in order to prevent high signal from causing grid current, but under 3.1V to keep the plate's voltage reasonable. However, LEDs in that range have a fairly high impedance, which drives up the plate impedance, which is a bad thing. So, on a suggestion from Colin, I am thinking of using a PNP transistor in an emitter follower configuration as a bias element. This is in essence like the pass transistor in a shunt regulator. The idea is that when the base is biased up some amount, Vce (the voltage across the transistor) is the bias + 0.7V. Plus, the impedance is very low -- just a few ohms. Here's what it looks like. Notice that the bias is set by an LED, and the LED gets its biasing current from the resistor string that provides biasing current to the CCS LEDs, so the current is recycled. If you want to implement this on your prototype boards, and I think you should as I think it sound a lot better (and it measures better), here's how: - jumper the transformers to ground (LED connection). - jumper for resistor bias. - remove the bias resistors (R3L and R3R), and replace them with PNP transistors (BC560, etc.) with the emitter connected to the trace that connects to the tube's cathode and the collector grounded. - desolder the ground side of R5L and R5R and stand them straight up. - run a small jumper wire (per side) to connect the unconnected side of R5*, the transistor's base, and the anode of the biasing LED together - For extra credit, connect the positive leg of an electrolytic cap (470u-1000u, 6.3V or higher) to the jumper wire with the negative lead to ground (the base of the transistor, or the cathode of the LED). It seemed to measure a tiny bit better doing this, though this is inconclusive. The overall bias will be the LED's Vf + 0.7V. I used a 2V greed LED for ~2.7V of bias. If you have red LEDs in place, they are fine, too.

Any chance you have a 120:120 isolation transformer around to use before this one? That would allow us to be sure it isn't DC on the power line. At any rate, it sounds like it is likely a bad power transformer -- probably a gap in the lams. If that's what it turns out to be, I have another I can send you. If you can't get that one off the board, I can do that for you, too. And be careful where you put your ear

Looks nice (and looks correct). Any impressions of resistor versus LED?

Are you 100% sure it's the transformer and not the choke? Hammond stuff is notorious for buzzing, but I've never heard an Edcor do it. If you can eliminate the choke as the culprite (remove it and replace it with a resistor), then try the amp without tubes and see if it still buzzes. Also, did you install the jumpers from A to B, and from C to D?

Does anyone have any idea of where to find some of these, or something else that would work for the same purpose, but with threads for screwing them to a bottom plate from the bottom?

And search around for quiet drives -- many hard drives are too noisy.

Look at post 361.

The cable in the picture shows what looks like non-fire related damage -- look at the cut at the base of the burn, and the blue a bit further down. I vote for run over by a chair.

An LED on the cathode is, in theory, similar to a resistor bypassed by a capacitor. At a fixed DC (as from a CCS), it drops a fixed voltage, but it acts as a short circuit to AC. Thus, it biases the tube, but does not increase it's impedance. An unbypassed resistor increases the tubes impedance by ~the resistance X mu. With the LED bias, since there is an AC short to ground, the signal current loop is completed through ground, and thus the transformer primary can be grounded. With the resistor, since there is no capacitor bypass, we need to short circuit the AC somewhere else. Thus, the primary is connected to the cathode which keeps the AC from traversing the resistor, instead being confined to the tube + transformer primary (remember that the plate and cathode are out of phase w/r/t/ each other). This arrangement seems to work better with high current and a high bias -- i.e., with a large resistor, as this is better able to block the AC. It is kind of like a cathodyne with unbalanced loads, and since the plate Z is so high (nearly infinite) you don't lose amplification even as the bias resistor is increased. With the small bias resistors we started with, I thought is was not great sounding. At higher a higher bias point, i.e., a larger bias resistor, it might be more viable -- I have not tried it yet. This does raise the question of why not not connect the primary to the cathode with the blue LED that has a pretty high impedance. And, the answer is that it seems to be not high enough and seems to muck with the phase. Try it. I didn't like it. It also raises the question of why not just use a resistor bypassed by a big cap -- there is some internet lore about how the distortion from electrolytic caps is not measurable, and this turns out to be nonsense. Especially in this spot. The distortion is multiplied by mu, and is clearly higher than the distortion even from a lousy LED. At any rate, I currently have 66.5R CCS resistors (15mA) and crappy blue LEDs, and it seems to sound pretty good at reasonable listening levels. Problem is that B+ is ~205V and plate voltages are 170V and 180V meaning that I don't have a lot of swing room. But, I can probably increase the power transformer by ~30V which should solve that ... at the cost of some more heat.

Actually, I think I'd push the CCS to more like 16-18mA or so, which means a CCS resistor of about 62-56 ohms. Plus, the bias should be over 2V, and closer to 3V if possible. Whether you can do 3.5V bias depends on how loud you listen, and where your tubes bias up.

The grid is at ground, and is referenced via the pot. The 1M is a backup reference in case the pot has a flaw. The cathode is biased at somewhere between 1.5V and 3.5V with higher being better but also needing to be balanced against plate voltage.

I think Pete's box is really only necessary for power amp testing as a way to protect your sound card.

I think that's enough of a difference to matter. But, try it and see what you think. Yeah, the impedance curve looks great on these. Um, not sure I'd go that far. But, I do think it compares OK at the anticipated price point. Don't think so. They are surplus. There are, but there aren't any more transformers. Send me an email ... My anxiety level has increased. You should find some Grados. OK, so here's the status. Tom and I are in a bit of a holding pattern trying to decide what the best next step is. Option 1 is to basically produce the amp as is. We would make some changes in the board and case designs, etc., but the circuit would be about the same. This wasn't really an option until playing a bit more with the bias and current, but with the improvements there, it might be viable. Option 2 is to basically leave the amp as is, but switch to a different tube -- probably a 6DJ8 with both side in parallel. These are obviously more expensive, though a pair of JJs is only $25 which isn't too bad. It would probably make a nicer amp, though the high Gm could lead to issues if builders are not careful, and it leads to the whole tube rolling nonsense, and I am not sure I want to be responsible for some idiot spending $400 on a pair of tubes. This is supposed to be a cheap intro to tubes. Option 3 is to make a little daughter board for different tubes, so one could build the default 6J6 version, or use a different board for 6DJ8's or whatever, etc. Again, this drives up the price for everyone, and not everyone would want it. And, option 4 is to alter the circuit using something like a Beta follower with a 6DJ8 again (6J6 won't work well). Upside is that it is a neat circuit that performs very well and it has a pretty low Z out meaning that the transformer ratio can be lower which translates to more power/volume. Downside is that it is a lot more complicated to get right for newbies requiring some adjustment of the HV section, and the parafeed cap would need to be a lot bigger. So, that's what our thinking is, and any comments from prototypers would be quite helpful in sorting it out.

Reviews have been something less than extraordinary New Wood Chassis from Hammond - Page 3 - diyAudio

If you put a small resistor in series which each, then they tend to do OK. [url=http://syclotron.com/?page_id=31]SYclotron Audio

Well, I think I spoke too soon. I picked up some 3.2V blue LEDs, and impedance seems to be in the 40 ohm range. So, maybe 2 red LEDs in series is better. Best might be 4 reds in series/parallel.

To put some numbers here, with the 6080, mu is 2 so gain of the cathode follower is ~2/3. Zout is ~150 ohms. With the 5998, mu is 5.5, so the cathode follower's gain is ~0.85 and Zout is ~75 ohms. Assuming that the gain stage is a grounded cathode with a CCS plate load, you should get a gain of ~20 from the 12AU7. With a 1V input signal, the 6080's output is ~13.3V into an infinite load, and thus puts about 8.9V across a 300 ohm load. In the case of the 5998, it puts 16.9V into an infinite load, which is about 13.5V into 300 ohms. The difference is ~3.6dB.

The 75R Zout difference (1/Gm) between a 6080 and 5998 is probably swamped with all but low Z phones -- and note this difference has nothing to do with having a higher mu. The "gain" (mu/mu+1) difference is pretty small, too compared to the high gain of the input stage. The mu of 2 6080 has a -3.7dB gain as a cathode follower, while the mu of 5.5 5998 has a -1.45dB gain.

It's a cathode follower. Why would a tube with different mu matter?

The LED sees the current, but it knows nothing of voltage. If you supply 18mA via a current source, the LED will drop Vf and the CCS will drop the rest of the voltage. if you have a voltage source of 100V, the LED will burn up. Edit: By the way, you can think of the tube as an adjustable diode here. At a particular bias, it has a Vf and a fairly high dynamic impedance. And, as with the LED, the CCS will drop any voltage not dropped by the tube + LED -- maybe someone should try using a TL431 as a little sand tube .

This is SY's claim I think, and I believe even he has backed off it -- maybe he ran out of his old LEDs See the part about theory and practice. Bottlehead, with their CCS kits, used to include some "special" LEDs that had super low impedance and that they claimed were the only ones that really worked -- they no longer do. Good red LEDs theoretically have a reported impedance in the 8 ohm range. Bad ones are more like 12 ohms.* I think anything you find will be in the high range. Green and Blue and whatnot might be as high as 12 to 14 ohms, but they are still generally lower than 2 reds in series. This impedance is multiplied by mu and added to rp. mu here is 38 and rp is ~3500, maybe more like 3K at high current. So, we are adding maybe 500 ohms to that due to the LED bias. To my ear, and by measurements I've done, this is more than made up for as opposed to a resistor and a cap. But, importantly, the difference between adding 400R and 500R here is pretty insignificant. Oh, and while you could theoretically put two in || to lower the Z, my experience is that you need a pretty high current before you hit the linear part of the Z curve, and we don't have enough for two here. * this is my understanding from what others have claimed, and from looking at datasheets. I have not tested them directly ... probably should

You are not likely to blow anything up. You should have an email with the jumper information. If not, send me an email and I'll send it to you. There are 4 jumpers, the two closest to the tubes switch between connecting the cathode LED and Rk, while the two by the transformers switch between connecting the transformers' secondary returns to the cathode and connecting them to ground. For the WE connection, you want the resistor connected, and the transformers connected to the cathode. For the LED connection, the LEDs (D3L and D3R) should light up and the transformers should be grounded. Please note that these are not the CCS LEDs which you should leave alone.

Not sure I know what you mean? The current is the CCS current. I'd set it anywhere from 12mA to 18mA to experiment. Voltage is the Vf of the LED. In an ideal world, you'd pick an LED with the lowers dynamic impedance you can. In the real world, this information is rarely clear from the datasheets, and LEDs don't really differ that much, so worrying about it is likely not that important. You know the old saying, that in theory, theory and practice are the same, but in practice, ...