DIY Audio Projects Forum

I managed to find the time to take a stab at modeling the ODDWATT in PSPICE. I just got PSPICE about a month ago and I'm amazed at what it can do, although finding reliably accurate tube models is difficult.

Other than using a different transformer and substituting a 12AX7 model for the 6SN7, the biasing and the output look pretty good for a first draft.

The only strange thing I see is a rise above 10kHz. The responses are shown below. The lower line on the graph is the input flat at 0dB and the top two lines (overlapping) are the two plate AC voltages - they match perfectly in magnitude but are 180 degrees apart in phase. Everything is in voltage, not dB. I'll plot the log response in dB after I get some sleep. It's much flatter looking. The curve above actually isn't that bad when viewed on a dB scale.

The transformer specs are in the lower left corner. I wish I could find specs for Edcor and Hammond transformers.

If anybody sees what I'm doing wrong other than what's mentioned above any help or suggestions as always would be appreciated.

NOTE: Added dB scale plot of plate AC output with 1 volt RMS input. Response is down 2dB at 20Hz and up 3dB at approx 25kHz.

_________________
Stephen

Hi, Pretty good sim. You have now discovered the two biggest problems with SPICE and tubes. There is a shortage of really accurate specs on the tubes and that complicates the sim and second many commonly used parts like the transformers have no specs. Still, sims can be helpful. At least they will keep you from doing something really stupid things. Now on the question of the peak in response. It is almost certainly from the transformer. I have found that transformers have a self resonant frequency. The Edcors I used have this at between 50-75KHZ. Well above the audio band, but close enough that it can cause sonic issues. That is the primary reason for the grid stoppers and the NFB loop. Together they keep the above band response in check. I suspect that the Dyna transformer resonates at a lower frequency, but have none to check. Also the sim may not truly reflect what is happening (back to the problem with accurate input data). I use two different DSO, two signal generators, an HP distortion analyzer and bunches (good techie term there) of meters to verify the response of the stuff I design and post as projects. So while the sim may show a peak, it isn't there in real life. The response is within +/- 0.2 db from about 10 HZ to past 22K at 1 watt and nearly the same through full output. I would be interested in the actual sim you used as I have 2 SPICE programs and would like to tinker with it to see if I can get the sim to match the actual. PM me if you are willing to do this. BTW, the original design uses 6SL7s not 6SN7s and I have a variation that does use 12AX7s however the resistor values are different from the ones in the original project. I also used 180 ma on the KT88s.

Good listening
Bruce

_________________
Some of my DIY Tube Amplifier Projects:

• PoddWatt: 5751 / EL84 Push-Pull DIY Tube Amplifier Project
• OddBlocks: KT88 Push-Pull DIY Tube Amplifiers (Mono Blocks)
• ForeWatt: ECC802S (12AU7 / ECC82) DIY Tube Preamplifier Project
• GrooveWatt: Tube (Valve) RIAA Phono Preamplifier Project

Correction: That's a 6SL7 not a 6SN7. I'm confusing my tubes. I used a 12ax7 model for the 6SL7 in the simulation becasue it was the closest thing I could find.

I think the transformer is the biggest problem with the circuit model. Supposedly the tube models aren't that far off, but like you point out, there are no accurate transformer models for typical off the shelf output transformers.

It's possible to measure some of the transformer parameters if you want to go to the trouble, but I think I'll just build the amp instead.

_________________
Stephen

Hi, The 9 pin equal of the 6SL7 is the type 5751. There may be models out there for it, I don't know. The transformers are the biggest problem in sims and modeling in general. That is one reason I stick with ones that work well. I am sure other trannies will work in the circuit, but exactly how well is unpredictable. Edcor trannies may not be noted as "high end" parts, but the measured results and listening to the amps convinces me that they are right for these amps. There is perhaps a synergy going here. The sum is better than the parts. In any case the amps are excellent performers.

Good listening
Bruce

_________________
Some of my DIY Tube Amplifier Projects:

• PoddWatt: 5751 / EL84 Push-Pull DIY Tube Amplifier Project
• OddBlocks: KT88 Push-Pull DIY Tube Amplifiers (Mono Blocks)
• ForeWatt: ECC802S (12AU7 / ECC82) DIY Tube Preamplifier Project
• GrooveWatt: Tube (Valve) RIAA Phono Preamplifier Project

Researching circuit topologies I came across something called a "long-tailed pair" otherwise known as a Shcmitt phase inverter. One of the uses is to drive guitar amp output push pull stages.

Am I mistaken or is the topology of the Oddwatt output section essentially the same but with a transformer as the load instead of resistors and a CCS used for biasing?

Thanks!

_________________
Stephen

Hi, Good observation. They are not really the same, however there are similarities. The Oddwatt ouput stage uses a combination of direct signal input to the one tube and cathode drive for the second one. The key element is the CCS. It needs to be linear and accurate (fortunately the 317s are pretty good at this). This allows the voltage on the cathodes to swing up and down in a linear fashion. You can almost think of the stage as a single ended one (it will BTW work that way at half the idle current) and a slave. Without the CCS the stage is rather non-linear and while it does work, the sound is only so-so.

Good listening
Bruce

_________________
Some of my DIY Tube Amplifier Projects:

• PoddWatt: 5751 / EL84 Push-Pull DIY Tube Amplifier Project
• OddBlocks: KT88 Push-Pull DIY Tube Amplifiers (Mono Blocks)
• ForeWatt: ECC802S (12AU7 / ECC82) DIY Tube Preamplifier Project
• GrooveWatt: Tube (Valve) RIAA Phono Preamplifier Project

I was running my preamp pretty hot to drive my oddwatt, so I finally decided to make use of the two 9 pin sockets I had installed "just in case" and do something about it.

I bought a pair of what are supposedly NOS Tung-Sol 5687 dual triodes (from TD) a few months back. They have black plates and man they get hot. Now I remember one reason I love octal preamp tubes.

I created a SRPP preamp stage from each tube using 1K resistors I had laying around, bypassing the cathode of the lower triode. Using just a few parts on hand I almost literally "slapped together" this preamp section. It's amazing how quickly something can get done once there are sockets in a chassis with power available.

I saw resistor values from 470 to 1.2k for the exact same SRPP using this exact same tube and similar values for other tubes, so I went with 1K to run a little on the low side. I'm using a B+ of 150 volts and I calculate the bias to be about 1.3 mA which seems a little low.

What's a good value to use with this tube for bias current in an SRPP using a 150 volt supply? Is 150 enough to get this tube to sound good or should I go higher? I can go as high as I want to with 475 volts available. I'll be testing a new high voltage CCS used as a voltage regulator and buffered with a high voltage transistor to regulate the supply to the preamp tubes. This voltage regulator would be powerful enough to provide regulated voltage to the KT88 oddblock. I don't know if it's worth the trouble but it should be doable. For the preamp section I want ultra-clean power. I have no issues with power tube hum, just preamp hum.

It sounds pretty good with the new preamp and I can now play any source no matter how weak through the oddwatt with no preamp needed. It's still a little noisy but I'll have that fixed. It has however lost something. It's not quite as clear and transparent as it was. If I can't fix it I'll put it back the way it was. I suspect rebiasing the 5687 would be a good place to start. That and raising the supply voltage.

Any ideas on what would make this sound a little "dull" to my ears. The schematic is the same SRPP that Bruce uses with the 6SL7 and 12SL7 but the resistor from the top grid to the bottom plate is just wire, and the cathode resistor (which is also 1K, same as the top resistor) is not split up to allow application of feedback, with the resistor fully bypassed with 500uF.

It looks really cool now with a total compliment of 8 tubes but my camera is lost (misplaced )

I was also wondering, should I bring the feedback all the way to this new first stage since I could apply feedback to an unbypassed portion of the cathode resistor if i wanted to , or would it be better to let the preamp SRPP remain autonomous and operate simply as a stand alone signal booster? To put it in the loop or not?

Any suggestions other than replacing it with an octal tube are welcome, even replacing the tube with another 9 pin. I actually didn't really know what I was buying but read it was a good preamp tube. I don't want to compromise the sound of the amp in any way, it was sounding awesome, so whatever it takes it what it takes.

_________________
Stephen

Changed B+ to 200 volts, resistors to 100 ohm. That seems a little low to me but the sound got better - a lot better. And the 5687s are

I think the bias is about 7mA now. It's hard to believe I need even lower value resistors. Am I burning up these tubes? I don't trust my calculation.

_________________
Stephen

Hi I'll have to run some cad on the 5687s as it is a tube I have not used before. SRPP stages are both simple and difficult. One can be made from nearly any tubes. How well it will perform is quite another critter. I have found that without going to really odd gyrations that a sort of sweet spot exists in the B+ range of 150-225. The voltages above about 180 though will require attention to the heater to cathode ratings of the tubes and may require a floating system with a positive DC bias for best results (and less likelihood of sparks and arcs). 6SL7s and 5751s work well at 215 v and about 1-1.2 ma. 12AU7s and similar at 215 and 5-8 ma. 12AX7 from 160-220 v and about 1ma.

For a regulator, try a LR8N series SS high voltage IC from Supertex. Use the TO92 version as it is easier to wire in. The flat pack is OK , but the in line one is tiny. They can handle up to 10 ma at voltages up to 450! I use them to regulate about 250 B+ to 215 for the tubes. Use one per channel for best results in stereo projects. Watch out for the dissipation ratings when you use them the TO92 is only good for 750mw.

Good listening
Bruce

_________________
Some of my DIY Tube Amplifier Projects:

• PoddWatt: 5751 / EL84 Push-Pull DIY Tube Amplifier Project
• OddBlocks: KT88 Push-Pull DIY Tube Amplifiers (Mono Blocks)
• ForeWatt: ECC802S (12AU7 / ECC82) DIY Tube Preamplifier Project
• GrooveWatt: Tube (Valve) RIAA Phono Preamplifier Project

I ordered the TO-252 package of the LR8. It’s smaller but it can be heat sinked to a copper board and is rated at over 3 times the power capability (2.5 watts) of the TO-92 (0.75 watts), although it IS small. Not too small to work with though using reading glasses and tweezers. I have big copper PCB pads that are there to do nothing but dissipate the heat. I biased the heater supply at about 1/3 the B+ supply.

I always wondered why so many people use preamps instead of simply putting the preamp in the power amp chassis. I think I found one very good reason. My problem is noise. I can’t get rid of the hum and buzz. There’s so much garbage RFI inside the chassis that it’s almost if not impossible to put a low noise preamp in there. The power is clean as a whistle. It's not regulated but I've got tons of filtering on it and that's not the source of the noise. I did everything I could think of short of magnetic shielding for the preamp but no go. Too much hum and buzz from all the AC bouncing around insided the chassis I think.

I thought it would be easy to get rid of, but no such luck. If I can’t fix it I think I’ll need to move the preamp into a separate chassis and live with two boxes.

Incidentally I tried using a BUX85 high voltage transistor as an emitter follower on the LR8 voltage regulator and it appears to work well. It vastly increases the output current capacity. This could conceivably be used to supply enough regulated power for the entire amp. I haven’t tried it yet in my amp but it breadboards well at lower voltages. It should be possible to get up to about 435 volts regulated output using the LR8 regulator with the BUX85 transistor as a buffer. The BUX85 can supply up to 2000mA., but 500mA is a more realistic maximum since hfe is rather low (about 50). For greater output current a Darlington pair could be used for better current gain, but I haven’t tried that yet. I have no noise problems in the power section so I doubt it’s worth the trouble. I’m thinking that it might be possible to reduce the size and cost of the filtering caps if the supply were regulated, but that’s just speculation – again I’ve never actually tried it and although it might be a little less expensive it's more complexity.

_________________
Stephen