DIY Audio Projects Forum

300B SET Design Project
Page 2 of 53

Author:  rock4016 [ 30 Sep 2010, 07:54 ]
Post subject:  Re: 300B SET Design Project

Electra-print can wind you some silver "pin striped" transformers for your application would run about $600 for the pair. That presumes single secondary. Electra print can also provide your choke and power transformer and provide them alll in the same quote, with shipping.

That's how I would proceed when building an amp with your expensive tubes.

Author:  mwhouston [ 30 Sep 2010, 08:27 ]
Post subject:  Re: 300B SET Design Project

With regard to OPTs, which do you think is the best, ignoring price: Edcor, Electra-print. Hammond?

Author:  rock4016 [ 30 Sep 2010, 10:03 ]
Post subject:  Re: 300B SET Design Project

Ignoring price Electra-Print is the best of those 3.

Author:  azazello [ 30 Sep 2010, 10:54 ]
Post subject:  Re: 300B SET Design Project

One Electron UBT-3

Author:  mwhouston [ 30 Sep 2010, 16:18 ]
Post subject:  Re: 300B SET Design Project

rock4016 wrote:
Ignoring price Electra-Print is the best of those 3.

Electra-print are half the price of Hammonds for me.

Author:  Suncalc [ 30 Sep 2010, 23:42 ]
Post subject:  Re: 300B SET Design Project

Wow! :shock: Interesting by-play on this thread. I hope I didn't start any fights.

It thought it was about time to get down to the design of this amp. Now normally for this type of project one would start with the power stage and work backwards to the input. And even though some of the design parameters have been set by the requirements stated earlier, it is still most prudent to start at the end (so to speak). So the power stage design comes first.

Back in August we were having a discussion over on the SET - Single Ended Tube Amps thread about how it seems that some tubes just like to be driven hard. This appears to be a characteristic of the typical triode curve shape and where things start to distort. As such, I decided that I would apply the same philosophy to this amp. This means that I want to get up close to the plate dissipation limit and the plate current limits of the tube.

Now according to the data sheet, the 300B has a plate dissipation limit of 36W and a max plate voltage limit of 400v. The plate current is limited to 70mA in fixed bias mode and 100mA in cathode bias mode. The requirements for the amp clearly state that this should be an 8W class amp. It will very difficult to get this power level along with low distortion with a 70mA limit on DC plate current so this drives us to using cathode bias for this amp. This decision will have an impact on the power supply because we now need a B+ voltage which is considerably higher then would have been the case if we could use fixed bias. The first attached figure shows the above limits superimposed on the plate characteristics curves and indicates the are where we want the DC operating point to go.
Power Stage Design 1.jpg

Very often this is the stage where people like to start drawing load lines, but I take a slightly different approach. The bias point location is set by more then just the output transformer primary resistance and the plate voltage. It is primarily set by the choice of bias resistor. The best design in the world is no good if one can't by a resistor to properly bias the circuit. (I know there is always the option of combining resistors, but this just adds complexity and using this approach building giant "Franken-resistors" is largely unnecessary.) I like to take several standard 5% resistor values and plot cathode bias lines on the characteristics curves. I use 5% resistors because these values are generally available in higher power ratings and there are 1% resistors of very close values if higher precision is required. This is accomplished using the constant grid voltage curves and plotting the cathode bias lines between adjacent constant grid voltage curves along constant resistance contours. Confused yet? :? Let me give an example.

Looking at the 300B plate curves, if I choose a standard resistor of 750Ω, then I plot a line between the Ec=-60v curve and the -80v curve where the current level on each curve is the assumed grid voltage divided by the resistance. Therefore on the -60v curve the end of the line is at 60v/750Ω=80.0mA and on the -80v curve the end of the line is at 80v/750Ω=106.6mA. Simple, right? :up: The following diagram uses this exact approach to show bias operating points as a function of several different grid resistors. Once these contours are drawn, you know exactly where the bias points can fall based on the desired plate voltage.
Power Stage Design 2.jpg

Now is the time for some of those arbitrary design decisions. As stated above, I want to be close to the plate dissipation limit and I want to be up toward the plate current limit as well. However, I also want some leeway to move the plate voltage around a little bit because I haven't designed the power supply yet. The curves for 910Ω and 820Ω are too far from the dissipation curve unless I get very close to the voltage limit. So these are out. Further, the 680Ω curve doesn't give me very much flexibility in changing the plate voltage without busting the dissipation limit. So it's out as well. This leaves me with the 750Ω line and therefore this is my preliminary choice of cathode bias resistance. Simple.

The next major decision is the choice of bias point and output transformer. The choice of bias point was somewhat arbitrary on my part. Based on judgment and a little bit of design margin I chose a bias current of 95mA. About 5% below the max allowed and with enough wiggle room that I can play with the power supply voltage a little. (This corresponds to a DC bias voltage of ~71v.) The transformer selection process is really one of first determining roughly where the sweet spot of the tube is at this bias point. This is a balance of power (which is generally maximized by lowering the output transformer impedance) and distortion (which is normally minimized by raising the output transformer impedance). I am going to spare you the tedium and just tell you that I tried general values from 1.5KΩ to 4.5kΩ in 1000Ω (and a few 500Ω) steps. What I found is that the sweet spot for this design falls between about 3kΩ and 3.5kΩ.

Now, just like the bias resistors, the design really isn't much good if it requires a 3.156kΩ transformer that I can only get by custom winding one. What I need to do is look at some available transformers and see how they work. Now luckily for me (and Mark) we have some good input from all you folks on the forum. Four suggestions for output transformer manufacturers have been made; Edcor, Hammond, Electra-Print, and One Electron. The Elextra-Prinf will custom wind whatever we want so lets place that manufacturer aside for the moment. Edcor has a very nice CXSE transformer in a 3.2kΩ primary, Hammond has the very good 1630SEA in 3.5kΩ primary, and One-electron has the massive UBT-3 in 3kΩ primary. But how do we choose? :confused: The answer is to check each and see what the design data tells us. The following figure shows the results of the three different transformer primary impedance choices using the chosen bias point. As can be seen the results look very similar at first but there are some subtle differences.
Power Stage Design 3.jpg

The three different load lines all produce about 8W. The difference between 8W and 8.2W is 0.1dB and the difference between 7.5W and 8W is 0.28dB. In either case, not enough difference about which to really care. (Unless of course, the customer is hard over on having "at least" 8W. This was not indicated to me, so I'm using a little leeway in what I call an 8W class amp.) The distortion is a different story. The 3kΩ primary has 33% higher distortion then the 3.5kΩ primary. Normally this might be of some concern. However, these values are actually quite low and the Ds parameters are actually much closer. Also, max distortion is only seen at max power. And as sound pressure level increases, the level of distortion which can be perceived by the human ear goes down. As such, it is unlikely that most people could even perceive a difference in these distortion levels. The other thing to remember about this is that you won't be presenting three different amps to the customer to compare and contrast. You're only building one and that one needs to meet his requirements.

So the question remains, which output transformer to choose? This is another one of those instances where judgment comes into play. The design process is always one of trade offs. The Edcor transformer is a known quantity and they usually sound very good but the max allowed bias currents are not published and the primary resistance is unknown. The Hammond has a great reputation and produces the lowest distortion but is more expensive and also produces the lowest power. The One Electron produces the highest power and the very impressive specs are completely published (this is actually pretty important to the design process) but is massive enough that it may present build problems. And lets not forget the Electra-Print. These have a good reputation for quality and we can pick the impedance we want, but they are not very common among the DIY crowd.

Personally, given what I know about each transformer, I would probably chose the Edcor in the 3.2kΩ impedance. It's right in the middle of the sweet spot, it allows me to actually claim a true 8W output, and they are a known quality product. However, as most designers who are working for someone else know, sometimes the Boss can override your decisions. So since I designing for Mark not myself, I am going to recommend the Edcor CXSE25-8-3.2K and let Mark either approve the design choice, or select one of the others. The good thing about this is that all the choices should sound very good. We're really only discussing small differences here.

Well, so much for the power stage. Next time we'll discuss the driver stage and how to handle some of the other requirements.

Comments anyone?

Author:  Geek [ 01 Oct 2010, 00:24 ]
Post subject:  Re: 300B SET Design Project

I trust engineers over superstitionists ;) ... rs_TSE.htm

Bottom line, I won't use any OPT except Edcor unless my client demands else.


** edit **
Wow! :shock: Interesting by-play on this thread. I hope I didn't start any fights.

Hehehehe. Pick an audio forum... any one... say "300B" and you'll have phasers and lightsabres coming off belts and into hands all over should the diplomacy fail :nuke: :hot: :D

Author:  mwhouston [ 01 Oct 2010, 03:23 ]
Post subject:  Re: 300B SET Design Project

Geek wrote:
Hehehehe. Pick an audio forum... any one... say "300B" and you'll have phasers and lightsabres coming off belts and into hands all over should the diplomacy fail :nuke: :hot: :D

Funny how you don't get the same reaction when you say 2N3055!

Matt: Mate, so much effort so far and such a great explanation. After needling Bruce for some further comments about Edcors (which I also happen to like but I must think beyond "I") I'm more than happy to go with the Edcors.

What is critical for me is to keep the project rolling. Where a large and important chunk of the componentry comes from OS I have to keep thinking ahead for what I need to order. The TJ 300B/SEs and Treasure 6SN7s will come soon and I need trannis and silver inter-stage caps to follow close.

Power tranni specs would be nice and inter-stage cap values would help here. I will order the Edcors tonight.

Author:  azazello [ 01 Oct 2010, 15:16 ]
Post subject:  Re: 300B SET Design Project

For cathode bias I advice You 750 ohm/12Wt from Mills, the best resistor IMO for about 8 Euro......Of course for HiEnd :D ... =4&sort=1a
I used for my SE 2A3RCA and SE 6S4S.

Author:  m.saunby [ 01 Oct 2010, 16:12 ]
Post subject:  Re: 300B SET Design Project

As a user of old stock I've never had the opportunity to test new production valves on my AVO valve characteristic meter. How well do tubes like 300B conform to published curves? I know that NOS from Mullard, GEC, Brimar, etc. are generally pretty good, but I'd never expect an exact match to published curves.

What's wrong with a high quality wire-wound pot for setting cathode bias? Then you can have whatever value is best for the tube.

Page 2 of 53 All times are UTC - 6 hours [ DST ]
Powered by phpBB® Forum Software © phpBB Group