Jump to content
The Klipsch Audio Community

John Warren

  • Content Count

  • Joined

  • Last visited

  • Days Won


Everything posted by John Warren

  1. Spent Friday last listening to the Transcendar and Hammond output transformer versions of the amplifier. I have a well broken in set of Sylvania 7591s that I was using, swapping them from one to the other. Speakers are JBL L200s. I also revised the feedback on the Hammond to provide a bit smoother response above 50kHz. Here's the distortion curves for both channels with the Transcendar outputs and Sylvania tubes, virtually identical. There's a leveling off as the power approaches 8W RMS or so then there's a significant rise past 15W RMS. The plates can dissipate 19W RMS. Same tubes with the Hammond outputs. The Hammond is rated for 60W, a much larger core than the Transcendar which is a 22W transformer. And here's the amps How did the sound? Hard to discern a significant difference between the two but I did like the Hammond version better. The L200 isn't that good in the mid range and the bass is "tubby" but, that said, both amps sounded good to my 62 y/o ears. I was plugging a 24-bit CD player directly into the RCAs, no preamp. The distortion curves were taken after my listening so I wasn't entirely biased based on measurements.
  2. I'll be giving a pair of Lundahl output tramsformers a go in this amplifier. I'm curious to measure how a C-core transformer measures against the conventional units. The LL1620/PP has the correct primary impedance (6k Ohms), wattage (25W) and max plate current (150mA) into an 8 Ohm secondary load. There's a nice powder coated metal enclosure that can be purchased too. It's almost an exact replacement for the OEM version but, theoretically, a better version. It will require a third set of mounting holes be added to the chassis plate. The advantage of a C-core transformer is the magnetic field is well contained and lower stray flux, no sharp corners in the iron circuit. The coupling between the primary and secondary should be better. But can it be heard? and can it be measured in %THD? .
  3. All things are possible.
  4. Yes, he contacted me a few weeks ago asking if he could purchase it so I sent it away. I've serviced a couple of his tube amps in the past. I'm curious to get his take, he has many vintage tube amps including some rare pieces. I've have a second Beta built and parts for a third. PC boards and the machined chassis plate have a small price break at three pieces. The amp will produce full power into an 8Ohm load (~12VRMS) and <0.5%THD with an input signal of ~1.6VRMS which allows for DVDs, CDs, tuners are other "line level" outputs including pro hardware. With sensitive speakers the output would be painfully high with a direct plugin of CD player and listener a few feet away. It will require a preamp for a phono inputs. There's a separate level adjust for each channel and a stereo balance potentiometer. There's also a line level center-channel output RCA jack with level adjust too. I always wanted a center-channel in my setup and now I've have one. I have the preamp board on order but have yet to consider how it will be packaged. I'm less excited about tube preamps. I've designed solid state preamp boards with greater than 100dB distortion free output.
  5. Here's the beta board set. The boards are designed using a package that links the schematic to the layout. So you first layout the schematic and all the painful details associated with the size of the leads, component body, wattage, pin identification and then commence with the layout. Dual traces (top and redundant bottom trace) are dark shadowed in the photo. Here's closeups of each board
  6. Some performance data comparing the two designs. Here, the analyzer was calibrated and left alone, the amps then measured sequentially. Same tube types and manufacturers (JnJ) but the alpha tubes have more time on them so not entirely an apples 2 apples comparison. DC balance was checked and adjusted before the test. 1kHz %THD. Both channels in the beta are similar in performance. FFT of the alpha unit, both channels (CHA is top): FFT of the beta (CHA is top): So some differences with beta being a bit better but not dramatic differences. Both sound great and are fun to use.
  7. Getting back to the Scott clones, here are the two amplifiers, the one on the left is the first iteration (i.e. alpha) using the first set of PC boards I designed for the project. It uses the Hammond output transformers. The second iteration (i.e. the beta) is to the right. That guy uses nice wire to board connectors a bit better trace layouts especially with regard to handling traces that drain back to the c/t. It also uses the Heyboer Scott 299C replacement PS transformer and the Transcendar 299C replacement output transformers. From the photo below you can see these are not small. Here's the Beta with cage installed. It went out FedEx yesterday so let's see how well it survives the trip to Chicago! The chassis plate is 1/4"thick 6000 series Aluminum and all the fasteners are good quality.
  8. Thanks JC, I'm pleased to hear they're working well. When I was done with JC's 250 amps, they both measured well, low distortion and ruler flat bandwidth. After a couple hours of running hard and loud, both were cool. I was pleased when they went out the door. When JC hooked them up to his rig, he's initial impression was they were hard on his ears. Apparently a wee-bit of fiddling with the setup got them to behave. I'll post a few photos of the work I did on JCs amps in the solid state forum.
  9. Thank you. The original amp is a two prong whilst this thing is a three-prong so there's now a "chassis" ground and "0V reference" (i.e. the center-tap). On the early McIntosh solid state amps the small signals were handled using single conductor instrument cable with the shield drain lead connected to the chassis at both ends of the conductor, the lead break-out was connected to the shield braid using a small crimp clamp and then the entire end of the conductor insulated with shrink. I've followed that approach here, the shield drain is connected directly to the amp chassis. On the RCA input side of the amp, the ground sleeve is actually a 0V reference and is handled differently, it's connected directly to the PS transformer center-tap. That's what's going on here, the green leads are 0V references, the black whips are chassis ground connected. The point in all of this is to make the amp as quiet. My take is it worked.
  10. Your amp was the inspiration Mark!
  11. It's a stand alone "basic" amplifier with CHA and CHB input level adjusts, a Center channel out level adjust and stereo balance pot. I actually did design a preamp board. I can duplicate the entire amplifier, including the tone controls. I designed the preamp board and ordered all the parts for the build, just need to order the board (the entire LK-72A preamp is on a single board!).
  12. I'm pleased you find something of interest here! I'm using a CLIO 12 24-bit analyzer @192kHz sampling for distortion analysis. It's a high precision FFT analyzer with a wide range of input signals for analysis. I also have Rohde & Schwarz UPV, Krohn-Hite and Keithley analyzer. I can certainly provide FFT spectrum but find %THD to be a pretty good indicator of amplifier capability (as plotted above). And yes, it's a HH Scott LK-72A/299C copy and you won't find this on Ebay.
  13. CHA and CHB bandwidth with the Transcendar LK72/299C output transformers. Some improvements over the Hammond units. -3dB is about 40kHz for each, not bad for a 60 year old design. The plots look similar to the OEM responses too. Input signal is 200mV cross-correlated MLS. Level pots open wide. %THD at 1000Hz below for each channel. The uptick in the distortion curves is associated with oscillation pulses that occur in the output when the tubes approach the design limits.
  14. Being true to the design original has it's problems. There are two 100uuF Silver Mica caps in the OEM units and one is in the feedback circuit sourced from the output transformer secondary. The amp will oscillate using the Mica cap in the feedback circuit and cause the output tubs to red plate. A 100uuF MLCC however completely defeats the oscillations. It's the little yellow one in the photo below. An explanation may be in how the caps are constructed.
  15. I actually had a few hours today to finish the second build. Some small changes to the PC boards to make it a bit easier to build. This amp sounds different than the first build shown a few pages back. The output transformers are the Transcendar 299C replacements for the OEM transformers. These present a bit better load to the 7591s than does the Hammond (6000 Ohm v. 6600 Ohm, respectively). I also have a very low dollar 5AR4 in the power supply section, that's nice and bright when on. Bottom side below. The small signal leads are instrument cable, the shield drains (black leads) source to the chassis ground. The approach of attaching the drain using a seperate lead is similar to the McIntosh approach used in 60s and 70s. The green leads to the center-tap. What's very cool about this amp is that it has a center channel small signal out which I kept in the clone. The four pots are CHA level, center channel out level, CHB level and Stereo balance. I thought about putting all of this on a PC board but boards in small quantities boards are pricey. The ones used on this build are sourced from Sunstone Circuits, top quality supplier. I left the leads long on the PS transformer. If this was a production unit they'd be trimmed. The AC introduction module has a filter in it to eliminate line glitches and other crap. I'll post some of the measurements taken on this amp next day or so.
  16. Thank you all for the comments, I'm glad a few of you find it of interest. The first iteration of this thing, posted earlier in the thread, was to determine if the basic layout of the components produced a low noise, hum free amp that could operate reliably for 100s of hours of operation. Here, a few revisions were baked in but the overall layout works well so I didn't change too many things with this go. Completed the CHA and CHB boards. Next step is wiring it up. There are three returns back to the HV supply card (green leads below), each lead sinks return currents from CHA, CHB and bias supply star grounding traces back to the center-tap. The HV supply utilizes it's own star grounding as well. Every component with a return back to center-tap has its own return to the center-tap which took some time to layout. I'm a fan of the AMP Faston connectors, they make w-t-b connections simple to enable. Purist would prefer solder but I'm not into making life miserable for fly-shit. On these units, I'm using the Scott replacement output transformers from Transcendar so I should be able to use the same feedback circuits as the factory originals. That was not the case with the earlier build using the Hammonds. A tweak might still be necessary. I use a couple for ferrules depending on the connection. Here I used the this one ($120USD) made by Zoller + Frolich (Crimpit F 6 L).
  17. My take, Tektronix oscilloscopes of the mid 70s vintage were the last generation of instruments designed around discrete devices. When you peek into a vintage Tek scope you see pretty quickly they represent an incredibly high level of design sophistication. The engineering, the supply chain and the industry as a whole had matured to a point where they were not going to get much better without a significant engineering breakthrough. That "step change" was leveraging integrated circuits. The point however is that discrete device designs, given where the technology was, were very effective at solving the problems oscilloscopes were used to solve at the time. They worked and they worked well. When ICs were used, bandwidth went up, signal to noise went up and the improvements were obvious on the screen of the scope. Analogous to the Tek example, tube amp design, manufacturing and the tube supply chain were, by the late 50s, mature, there was no place to go. And, they were making very good sound. Then low cost transistors appeared. Distortion went down, signal to noise went up and the improvements were obvious to measure on FFT and distortion analyzers. What hasn't had a step change is the loudspeaker and the basic loudspeaker of 2020 isn't that much different than the ones made in the 50s. Materials are better and magnets can have insane BL-products for a given magnet volume but they're only evolutionary changes, not technological "pivots". Unlike the oscilloscope example the audio signal path "ends" using a device invented in the 1920s and refined largely to a point where it's a bit better than where it was in the late 1970s. So yes, it's easy to measure differences between transistor amps and tube amps on distortion analyzers loading each with a resistor load. But the end of the signal path is not a resistor, it's thing called a loudspeaker, an electromechanical device that has mass and presents a reactive load to the amplifier. And, it can take the most sophisticated, electrically engineered audio amplifier made today and send it back to 1975 or even 1955 if you're using an EV Patrician let's say. So that's why tube amps still "work".
  18. That's actually 100mA, so plenty to stop your heart.
  19. Getting back to my initial question. I suspect that the risk of lethal shock, though quite real at 500VDC, is in fact lower than what theory would suggest. In other words, most shock events at 500 VDC deviate sufficiently far enough from the conditions required to cause a fatality. Now 1kV and higher is a different kettle of fish.
  20. There's a very real possibility of lethal shocks working with or tweaking a tube amplifier. In HiFi, they've been around for over 70 years and, combined with musical instrument amps, 10s of millions (if not 100s of millions) manufactured and distributed throughout the world. And even today they're still popular and in wide use albeit amongst a niche group of audio and musician types. And then there's kits. You'd think that death by electrocution would be a running plague associated with tube amps but I don't see it and find that a bit curious. Thoughts, comments?
  21. Just use it and enjoy it. When it craps out then think about getting it fixed.
  22. Thanks. Took about 4 hours to build what's shown below including testing the HV outputs. There are five HV rails. To get the module good and hot, ran them into resistive loads for about an hour at 125VAC input. Draws about 1100mA. B+ at 125VAC is about 460VDC. Uses 5AR4 rectifier, which I have many.
  23. CHB output module is next up. I didn't get much further than getting the Fastons soldered in. Completed the -45VDC bias and preamp filament supply. The 299C and LK-72A both use a separate bias winding off the power supply transformer to provide AC voltage to the supply. That's nice.
  24. Continuing on with the second iteration of the amplifier. The high voltage board terminals are now 600V Phoenix type. A dedicated terminal block ties center-tap leads from the CHA and CHB output modules back to common junction which then ties to chassis ground and the IEC grounding lug located on the filtered AC intro module. AC filament power now comes from Keystone terminal lugs. The large cap is a screw terminal type that allows for easy field replacement.
  • Create New...