Output coupling cap and transformer mod...?

Those are Solen caps BTW. but 400V was probably chosen for extra voltage headroom as you don't want these caps to fail... ever. 

Solen caps jump from 250 to 400V ratings. You *could* get by with a slightly lower voltage rating, but reasonable practice is for the output cap to be sized to handle AT LEAST B+ so nothing lower than 300V under any conditions. It might open up a few options in electrolytic caps, but mneh. 

 

I'm not a big fan of how OTL amps work with low impedance headphones, so I would lean away. 

 

If you want to do it anyways you should be able to find electrolytic caps in the size (physical dimensions, and electric capacity) and voltage ratings you need.

I am biased towards just a straight electrolytic OVER a bypassed electrolytic - I find the bypass cap leads to a certain smearing of the sound. 

 

If you don't want to upgrade the caps, but still want to do something - You may have good results with adding parafeed output transformers or autoformer (600:32ohm, 1K:8ohm, etc), and just running them off of the high impedance output jacks with the caps you have now. 

Hammond makes a 600:8 transformer off the shelf. There are a few places that will custom wind at those impedances - a bit more expensive, but much nicer. 

 

As another option in para-feed land, some DIYers have used toroidal transformers as parafeed transformers. They make a very fun alternative to "real" transformers, and work much better than you may think. 

Edited by nikongod - 2/21/13 at 1:03pm

ESR does not matter for output caps. Consistent ESR does.

Assuming there is no feedback in the circuit, the Zout from a 5998 cathode follower is about 70 ohms. For the purposes of calculating the 3db point of a coupling cap, you add this to the load impedance. That means your 50u caps give about a 30Hz 3db point into 32 ohms. It is definitely going to be a trick to get a larger film cap in there. If it were me, I might try to just parallel a second 50u Solen cap in there. Alternatively, you could add some series resistance between the cathode and the cap (100R or so). There is already an impedance mismatch -- a resistor won't make it much worse.

Looking at it again, 5998's are dual triodes. I don't know the circuit, but if they put the two sides in parallel, then the Zout is more like 35 Ohms. Or, it could be a different circuit. At any rate, you are probably correct that 3db is lower than you thought, but it may not be as low as I thought.

 

I like the idea of just putting the transformers into their own little box, and have done the same when I did this for a Bottlehead CRACK. It lets you move them from amp to amp and saves the trouble of figuring out where to cram them in the bigger amp. 

 

There are a few ways to look at the weird primary/secondary impedance. 

 

To start with: The DV337 has a simple cathode follower output stage. Simple cathode followers don't require a load to work properly, so going from 600ohm to 2Kohm (or more) is not an issue. So don't worry here. 

 

There is a possible issue with the transformers. Because of how transformers are made transformers DO like to have a load. They are not as picky as certain speaker amps that can blow up if they are run without a speaker hooked up, but transformers often have a certain way they work best. 

 

There are 3 ways of looking at the problem of figuring out how they work best. In order from worst to best they are:

 

1: Forget it! Just act like its a non-issue and don't do anything! 32ohm speaker (or headphone) on an 8ohm transformer - no problem, act all casual and stuff... like nothing we do maters. This is BS, but it happens. 

 

2: Take the assumption that the MFR's recommendation for the secondary load is reasonably accurate and then calculate what resistor gives that load when placed in parallel with the headphones and just plug that in. Since that transformer has an 8ohm secondary we can assume that it wants an 8ohm load. 12||32~8ohms, nice and easy, and almost always better than #1. 

 

3: Measure with function generator and oscilloscope. Don't worry, I have not gone off the deep end into the madness of measurements first design BUT measuring can speed things up SO much and can save you a lot of frustration in trying to hear things that may not be there. Play a 10Khz square wave through the WHOLE system and measure what comes off the secondary on an oscilloscope. Try a bunch of loads on the secondary (I use a pot) and look for a nice square wave with no overshoot BUT also the steepest leading edge possible. 

 

So you know how I said that #1 was BS, sometimes it isn't! There are some transformers that don't really care what impedance loads they have on their secondary. Mysterious beasts... I am sure there is a trade off in their construction or performance but I don't know what it is. Measuring will tell you quite quickly how the transformers you have behave - which means that even when it doesn't matter if you measure or not, it matters. 

 

Secret option #4

Rough tune your gear using #2 or 3 and then adjust to taste by ear. 

Edited by nikongod - 2/22/13 at 8:40pm

To clarify a bit -- a cathode follower needs a load. It is provided by the resistor between the cathode and ground in parallel with whatever is being driven -- the phones, the next stage, etc. If you remove your phones, then it is just the resistor. At a minimum for good performance you really want it to be at least 3x rp of the tube. The higher the better, which is why people put CCSes under cathode followers. It is also the reason that these big cathode follower amps have high distortion -- the load provided by the phones is too small. So, driving a transformer will likely yield better results than driving the phones directly.

When designing transformers, there are several tradeoffs. To overly simplify, for a given core size made from a given material (steel, or nickel, or whatever), each loop of wire can handle a certain amount of voltage (before saturation) and adds a certain amount of inductance. So, as you add more turns, the voltage handling increases as does the inductance. But so does the DCR and capacitance. So, if you want more power handling, you can increase the core, or increase the number of turns, with drawbacks to each strategy.

For the 119DA's, Hammond designed them for connecting 8 Ohm speakers to amplifiers that expect to see a 600 Ohm speaker. In the design process, in order to increase power handling, they had the option to make the core physically large (which is expensive) or add more wire turns, which is cheap. I'll let you guess which they did. This increased the inductance, which means that while they are rated at 600:8, the reality is that they are much higher impedance (the impedance is 2piFL where F is the frequency and L is the inductance). The benefit to us is that they can reflect a much higher load than 8 Ohms. So connecting your phones across the secondaries without any additional load is fine. You can add a resistor if you like but it should not be necessary.

Last, if you do use these, the 50u caps are larger than they need to be. With Grados, consider them to be 2400 Ohms and size the caps accordingly. Or don't. The 50u's will be fine. But you can shrink them if you like.

 

This will not cause ground loop issues except in the most unusual of cases. 

Specifically, if everything is sitting on a grounded surface (desk or metal filing cabinet) and the chassis of the enclosures ground through that. 

The odds of this happening are really low. 

 

The load resistor on the TRS jack is there so you don't get annoying pops when you plug the headphones in. DO NOT REMOVE IT!

How many ohms is it? 

2Kohm or more is no problem. 

If it's less than 2Kohm replace it with 4.7Kohm.