Back to home page

Homemade HiFi Amplifier


I originally wanted to build an amplifier to go with an antique Meissner AM High Fidelity tuner that I had restored. But by the time the amp was finished it sounded so good that it has become my amp of choice for listening to CDs while working in the shop.

The only parts purchased new were the resistors, capacitors and an aluminum chassis from Hammond. I made use of parts that had accumulated from parted out chassis and Ebay auctions. I estimate that there is around $50.00 total invested with about a week of building, rebuilding, tweaking and experimentation. The output transformer is from a Hammond church organ chassis from Ebay that also netted two 6L6 tubes, two 6SL7s and a couple of 6SN7s. I got the 5V4 from Ebay for $6.00. The power transformer also came from a lucky Ebay win in which I got two transformers for $10.00.


Input and Phase Inverter Stage

The signal from one or both input jacks is combined with a resistor network consisting of two 100K resistors and a 50K gain control pot. The signal is amplified by the 6SL7 twin high mu triode connected with elements in parallel for low noise and to provide a feedback point for the global feedback. The stage has very little gain due to there being no bypass capacitor around the cathode resistor, global feedback and local feedback through the 470K resistor from the plate of the input triode of the Schmidt phase inverter.

The output is passed to the grid of triode one of the Schmidt phase inverter by a .005 uF capacitor. I found that any larger coupling capacitor here produced distortion by a phenomenon known as 'blocking'.

The 50K resistor to ground in the cathode of the phase inverter is large enough to provide a 'constant current' source and allows the plate resistors to be the same value. Any value larger than 50K raises the cathode voltage above the heater-cathode rating of the 6SL7. The output signals are virtually the same amplitude but inverted in phase. I could detect no distortion on the scope at the plates of the inverter with a .5 volt input signal. (For testing, I used a Heathkit sine/square wave generator that I have restored.) However, I added local feedback by using a 470K resistor between the first phase inverter plate and the input amplifier plate.

Back to Top

Output Stage

The output is two metal 6L6 beam power tubes connected in a push pull arrangement. Fixed bias of -24V is applied to the grids through the 100K grid resistors. This a little more bias than the RCA tube manual calls for but the tubes run cooler and use less power with no signal applied. I found very little or no distortion was added by upping the bias by a couple of volts. The tubes are biased for class AB1 which means there may be cutoff on peak negative grid voltage excursions but no grid current may flow during peak positive grid voltage excursions.

I found that the two .003 capacitors were needed for stability. I first tried local feedback from the plate of the 6L6s to the plates of the 6SL7 phase inverter using a 470K resistor but found the capacitors provided better stability and lowered the gain for frequencies out of audio range. The output transformer came from a parted out Hammond church organ that had two 6L6s in push pull. (The chassis had a 6L6GC and a 6L6GB installed when I got it.) I don't really know what the impedance is but figured it should work since it was originally used with 6L6s.

I had planned to not use global feedback but experimented a little (really a lot:)) and found some feedback network values that produced a flat response with no stability problems. For testing I connected a five ohm 25 watt resistor to the speaker terminals. With the feedback values in the schematic and with a resistive load the response is FLAT from 20 Hz to 20K Hz at 18 watt output. With an RCA eight-ohm center channel speaker consisting of two bass/midrange speakers and a ribbon tweeter and an eight ohm sub woofer connected the response rises about 3db at 35 Hz, drops to nominal at 100hz and is flat up to 18 kHz. This produces a somewhat 'bassy' sound but is not objectionable. In fact, it sounds GREAT. A square wave is reproduced faithfully from 60hz to around 10Khz with very slight overshoot that damps in one cycle. Below 60hz there is increasing distortion but the signal is recognizable as a square wave. Above 10Khz the output gradually becomes a sine wave.

Back to Top

Power Supply

The power for the amplifier is provided by a pretty much standard full wave rectified power supply. The 5V4 is a heated cathode tube that provides a 'soft start' due to the slower heating of the cathode. I added a 6.3V 150ma lamp in the center tap to ground connection to provide protection in case of shorts in the B+. It is kind of neat to watch the lamp brighten when the volume is turned up. There is also a 1.5 amp fast blow fuse in the transformer primary.

The bias supply is copied from an old RCA tube manual sample circuit. The choke was made from an old radio power transformer that had an open secondary HT winding. I removed all except the primary winding and it works fine as a filter choke. It has a resistance of about 50 ohms but I did not measure the inductance. I figure it is in the two to four Henries range. There is about six volts peak to peak of ripple measured at the 100uF cap terminal. This is OK since the ripple is canceled in the push pull output as long as the tubes are fairly closely matched. (I cannot hear ANY hum with my ear touching the speaker and volume at maximum.)

The 6L6 screen voltage is provided by a voltage divider and filtered by a 47uF capacitor and .22 uF capacitor. The voltage divider arrangement provides some regulation of the screen voltage as the screen current varies greatly from no signal to maximum signal. The divider also discharges the filter capacitors quickly when the unit is powered off. This is a safety issue when working on the amplifier.


This was my first attempt to build an amplifier since I was in the eighth grade in 1955. I built a two-tube amp then from plans in an electronics magazine. It didn't work. I discovered there were glaring errors in the instructions. I wrote the editors and they responded with 'sorry about that. Deadlines and such you know.’ I did get it working and used it to modulate a homemade flea power transmitter by connecting two output transformers back to back.

Other Projects

Update I have since built a couple of similar amps.

Here is one of my efforts building a 'Part 15' transmitter. It would probably make a better Ham QRP rig for 160 though. Needs a crystal oscillator for stability.

Heathkit DX-60 Audio Mods

Here are my changes for the Heathkit DX60 audio circuits to alleviate the 'muddy sound' for which the transmitter is famous. All of the changes were made with the goal of broadening the frequency response and reducing distortion. Lowering the value of the 12AX7 plate resistors reduces the effect of load capacitance on high frequency response. The one meg gain pot reduces distortion caused by AC loading and increases the gain. With the added gain I was able to put in some heavy feedback. The negative feedback from cathode of the 6DE7 to cathode of the second section of the 12AX7 lowers the impedance of the driver and thus reduces the distortion caused by the extreme variation of load. The load varies from no load at all when the screen is driven negative to fairly large when the screen is driven to the positive peak. There is no rocket science involved. The back of an RCA receiving tube manual from the fifties or sixties or a 'Radio Amateur's Handbook' from the same period explains most of the theory. Heath engineers were designing a circuit to cut the lows and the highs to narrow the bandwidth of the transmitted signal. My voice has very little content below 100 Hz and above 3000 so there is little chance of splatter just from a too wide source. It just sounds better and more 'natural'. I tried adding a 3300 ohm cathode resistor in the first 12AX7 cathode for 'proper' cathode bias. It doubled the AC hum probably due to the increased impedance from cathode to filament. The 'contact/grid leak' bias is fine for the signal levels involved anyway. The filament really needs to be raised positive in relation to the cathode for hum reduction but can't be done here due to one side of the filament supply being grounded. The over modulation clamp just cuts off the drive at near cutoff of the 6146 leaving a residual RF signal. A capacitor across the zener might further reduce the sharp change in wave shape but the RF is already at minimum so the splatter and distortion caused by the sharp change in slope should be minimal. Just my thoughts anyway. I used a 22.5 volt zener but different 6146 characteristics might call for a lower or greater value. I arrived at that choice by monitoring the RF out with a scope. Yes, I did enlarge the hole in the panel to accommodate the new mike jack. I used a chassis punch. I also drilled a small hole for the modulation LED. Not shown on the schematic is the PTT. I just stuck a golf tee in the CW key jack to open the contact and connected a wire from the PTT on the mike jack to the key jack and use the normal CW keying for PTT. Someone somewhere has probably done it all before. It's hard to to come up with original modifications on a nearly fifty year old piece of ham equipment.

Heathkit SB-200 Mods

Here is a schematic of my SB-200. I made a few changes. The original 125uF filter capacitors in the HV power supply were replaced with modern 220uF 105 Degree capacitors. The 27K bleeder resistors were replaced with 82K three watt resistors. A 50 ohm 25 watt glitch resistor was added in the HV lead. A keying circuit was added to protect any modern transistorized rig from the 120 volts bias in the SB-200. A FOD852 optocoupler and a three volt power supply was a simple addition. Note the two diodes in the keying output from the FOD852. The diodes serve to isolate the FOD842 from possible tube shorts and also provide the -2V bias during key down due to the combined forward drop across the FOD852 and the two diodes. A diode was added in series with the 2K resistor across the keying relay. The diode eliminates 70 ma of current through the FOD852 and still allows normal transient protection. The 2K resistor is no longer needed to develope the key down bias.

  Any statements made or opinions expressed are purely my own.

All schematics were drawn using Tubepad and Microsoft Paint. Tubepad may be downloaded here.

Back to Top