ATMega8-Powered Peltier Controller


In this project a set of four Peltier coolers, also called TE Coolers, are to be controlled via USB. The power requirements for the coolers are high, +/- 12 volts at up to 7.5 amps each. The ATMega8 on this MPU schematic page is responsible for controlling duty cycle and current flow direction through an H-bridge circuit, which is on another schematic sheet. This control is accomplished via a set of commands issued to the ATMega8 from an application running on the host PC. There is also a thermistor and A/D converter connected to the ATMega8, which will report the temperature the TE coolers have achieved at the target. The completed unit will require firmware that is not too complex, but could grow to include a full PID (Proportional-Integral-Derivative) algorithm. This will be programmed in C via the AVRStudio IDE.

Series Regulated all-Tube Power Supply


The Kay model 397 power supply, a surplus store find. Full of some good parts, this actually works, and I’m considering how to use it in a  project as-is. It was originally used in a television-related test instrument, but was found all alone. I’ve replaced the power cord, built a new interconnect cable, and replaced some small capacitors underneath. This would have been part of a very expensive piece of equipment back in 1952 when it was built. I have a freshly drawn schematic and brief description for you:

Kay Power Supply_Bill

Here is the schematic of the Kay  Electric Company power supply. This circuit is typical of vacuum tube era regulated power supplies, with 6Y6 pass tubes, a 6SL7 “long-tail-pair” or current mirror voltage amplifier, 6J5 error amplifier, and VR150 150-volt reference tube. Note how this circuit is supplied by a floating 6.3VAC filament supply, while a separate transformer supplies 6.3VAC to the off-chassis circuitry. This second supply could be biased to eliminate hum in that circuit without exceeding heater-cathode voltage limits in this circuit. The output is adjustable via the potentiometer next to the 6J5, and with the components shown, the range of +250 to +300 is accurate. Changing the 82K resistors in series with that potentiometer would increase and/or alter the adjustment range. This power supply might be useful as a bench test instrument for trying amplifier circuit ideas, or to allow construction of finished items without each having an included power supply. What it does not include as-is is any sort of negative bias supply, although one could be added very easily.

Single-Ended 6B4G Amplifier

DSCF0473 - Best

This is my current favorite and latest build. This is the cure for any single-ended amplifier addiction. It is built on a 17″ x 10″ hammertone-painted steel chassis, and features Transcendar output transformers and an Edcor power transformer. This is the first directly-heated triode amplifier I built for myself, and had the time to wrap my ears around. It’s an amazing synergistic match to my sealed-enclosure speakers, and sounds outstanding at moderate listening levels.


This is as simple as such an amplifier can be, and I always believed shorter is sweeter when it comes to audio signal paths. The two 6SN7 triode sections are directly coupled, with one capacitor coupling to the output tube. All stages are cathode-biased and bypassed to ground. A worthy experiment might be to connect the output tube cathode bypass cap from the cathode up to B+, a connection often referred to as ‘ultrapath’. The power supply is very ordinary, but has a high inductance high current choke to keep hum down. The heaters and filaments are all supplied a rectified and filtered 6.3 VDC. Hum balance potentiometers are used across the output tube filaments so any hum there can be adjusted out at that point. See the global feedback? Nor do I! There isn’t any! This runs open-loop. The large output triode should help to keep output impedance down, but it will still be somewhat high, so damping will be poor. This requires the right sort of speakers to use effectively. When it does work out, it has punch, bounce, and puts you right there with those musicians. The bad news? This amplifier will reveal all of your poorly recorded music, and change your musical tastes to well recorded music.

Single-Ended 1626 “Darling” Amplifier


This is an amplifier type often referred to as the “Darling”. It uses a WWII-era oscillator triode, the 1626, as an output tube. This is all about finesse, as the power output of this tube is only 3/4 of a watt! This is built on a hammertone-finished steel chassis, and features volume control and headphone jack. This was built as a gift for my wife, and she uses it in her office with mini speakers, or headphones, and her IPod as source.


Here is the schematic for the 1626 SE amp. A three-stage CLCRC power supply filter was needed to reduce hum to inaudible for headphone use. A voltage doubler supplying DC to all filaments was used for the same reason. Otherwise a very normal audio signal path. The 6J4/8532 is an interesting tube to use here. It is one triode per tube, and often microphonic, meaning it picks up mechanical vibration and amplifies it. This means I had to try several before finding two good ones. The 8532 version of the 6J4 is usually better. Another feature of this design is that the two channels share the cathode resistors & bypass capacitors of each stage. Although output is an amazing 3/4 of a watt, it sounds very nice on headphones and speakers that are sufficiently efficient.

1626 6J4

Low Voltage Single-Ended Madness!


This little amplifier was inspired by a simple design found on the ‘net. I had found some Philco output transformers from a stereo record player that had tapped secondaries, which measured to be 2500:8 ohms or 4000:8 ohms. This opened up some interesting possibilities for using different output tubes in the same circuit but maintaining good impedance matching. If a 6Y6 is used, the cathode bias settles out at a current level that pulls B+ to 205V. Connecting the speakers to the 2500:8 taps gives the best impedance match. If a 6W6 is used, the current draw will settle out lower, pulling B+ down to 230V. This is where 4000:8 is the best match. The amp was built on a hammertone finish-painted steel chassis, with the power transformer under the diecast aluminum cover, and the output transformers under the chassis. This will output 3.5 watts per channel with 6Y6, and 3 watts per channel with 6W6.


Here is the schematic of the Madness, which is actually my first SE design. Overall similar to the previously posted 6Y6 amplifier, there are several important differences. This is connected as single-ended pentode, which will deliver more output but requires some sort of global feedback. It is also cathode-biased, which lowers output but simplifies design. The power supply is CRC filtered solid state, with a large R (680 ohms). This gives the power supply a lot of what is termed as “sag”, where the output voltage varies greatly with load. The interesting effect of this power supply is that the amplifier never ‘clips’ when overdriven, it simply compresses! This, like several of my projects, was a budget build that almost shouldn’t work out, but with a little tuning, it worked out beautifully.

Single-Ended Loktal amplifier


This interesting little amplifier started life intending to be a loktal version of the 6Y6 amplifier already posted. I did not get the results I was hoping for with it, so it was redesigned. Here you see the amplifier built on a Hammond powder coated black chassis. In front are two 7B4 tubes as voltage amplifiers, and behind are two 7A5 tubes as power output. The design will take 7B5 and 7C5 just as well. The transformer cover is actually a pre-painted diecast aluminum box also made by Hammond.


Here you see the design of the finished amplifier. The power supply is solid state with heavy filtering to keep hum from the b+ supply. The voltage supply to the 7B4 tubes and the output tube screen grid voltage supply are not stacked, allowing the 7B4 plate voltage to be higher. This gives some improvement in gain and linearity. The output tubes can be any of 7A5, 7B5, or 7C5. All three have the same base pinout. Although the three tubes have different operating characteristics, this circuit will allow their use. The output tube cathode bypass capacitors are not routed to ground, but rather to B+. This is a connection called “ultra-path”, and in theory should help isolate signal AC from the power supply, basically removing the power supply from the signal path. The output transformers were originally radio-grade 5000:8 ohm parts, but were replaced by higher-quality transformers from a vintage reel-to-reel tape deck. Because this is a single-ended pentode design, inverse feedback is unavoidable or the output impedance will be very high. This makes an amplifier sound funny, and that “funny” will vary depending on the speakers used! The feedback here was carefully tuned on my speakers by playing music and turning a potentiometer connected in the loop. The human ear is still the best test instrument for audio! As long as you haven’t played it too loud….

Push-Pull 5881/6L6 Amplifier


This was the second amplifier I built for myself and was completed in September 2006. It uses a power transformer scavenged from a Fisher 400 Receiver and Hammond 1645 output transformers, and produces 25 watts per channel. It was originally planned for the Fisher 400’s output transformers, but one turned out to be failed. Because the Hammond transformers have tapped primaries for distributed load operation, the output was rearranged for this topology. The input and driver stages are almost the same as the 6V6 push-pull previously posted because the former was based on this design, which is itself based on designs used in Pilot models AA-410, AA-902, AA-902A, and others. All tubes in this amplifier are of course re-selected to suit the builder’s tastes. Soviet-era 5881s and loktal-based 7N7s branded “Standard Tuner” bring the music. I do use a lot of loktals, because I claimed my stash before they got popular.


This schematic looks a lot like the 6V6 push-pull amplifier, but is just bigger and more powerful. The power supply mirrors that of the Fisher 400 receiver, as that was the source for the power transformer. Heavy capacitive filtering eliminated the need for a series choke, and the output is hum-free. The completed amplifier has a powerful, clean sound which is unlike the Fisher receivers, the Dynaco amplifiers, and others I have heard using split-load phase inverters. This may be because those items did not use the magic combination of all-triode operation and distributed load topology. This sort of design is very easy to lay out and wire, and gets excellent results.

Push-Pull 6V6 Amplifier


This amplifier was a gift to my father-in-law for Christmas a few years back. It is a distributed load push-pull output topology called “Ultra-Linear” back in the day. That is driven by a direct-coupled split-load phase inverter, preceded by a single gain stage, all using the common 12AU7. With this circuit no preamplifier would be needed, as my father-in-law has a turntable with line-level output.


This schematic represents possibly the best bang-for-your-buck project I’ve done. Although parts were purchased new to ensure a “shiny & new” appearance, they were not expensive. This circuit contains all of the features that make tube amps what they are while minimizing inherent drawbacks.

An Altium PCB Project: Design, Layout, Prototype


This project was my first using Altium Designer instead of Eagle or PCB123, the free packages. I helped to design the circuit, enter the schematic, create the PCB layout, and get prototype boards manufactured. In each phase I used Altium myself to perform a substantial portion of the design. Here is the ‘top’ schematic page, in which some symbols represent other pages. The MPU, an Atmel component, is shown. This design was scrapped and no code was written for it.


Here’s a screenshot of the prototype PCB’s layout. All portions of the layout, including the portion I did, were done by hand. Autorouting was not used. I sent the output (“gerber” files) to a local PCB manufacturer and ordered a couple prototype boards.


This is the prototype PCB that I hand assembled. I did this with .020 lead-free solder, a very fine Metcal tip, and a binocular microscope. The missing component was an expensive one that wasn’t necessary for basic testing. The power supply section was tested, and a programming connection via the AVRStudio IDE was made to the Atmel MPU. The project was revised before any code was written.

Single-Ended 6Y6 Amplifier


This little amp is the true junk box bash. Nothing about it was expensive, or optimal. The 6Y6 output tubes are triode connected for an amazing 1.7 watts per channel! No feedback loop, I had no gain to spare. the 5965/12AV7 was used because at this low voltage, it’s high transconductance gave adequate gain. a 12AT7 would have been second choice.


This is the final schematic of the piece. As you can see, the power supply is far more complex than the audio circuit. With the two-stage all-triode arrangement, and no feedback, this amp really cares what speakers it is connected to. It sounds best on well designed acoustic suspension speakers that have some damping of their own.