The Heathkit A-8A is a rare Heathkit integrated amplifier capable of around 15 watts per channel. It has three inputs — crystal phono (which can be used with line-level sources), mic, and magnetic phono. The tube complement is two 6SJ7 pentodes, a 6SN7 double triode, two 6L6 beam tetrodes, and a 5U4 full-wave rectifier. I bought this amplifier from a Craigslist seller who intended to restore it but never found the time. This example has a Chicago PCC105 power transformer and a Peerless S516A output transformer, but other A-8As may utilize different iron; the schematic and parts list don’t specify these models but rather provide Heathkit part numbers.
A Mallory multi-section can capacitor is mounted atop the chassis; despite the use of an insulating phenolic wafer, its shell is at chassis potential and should not present a shock hazard under normal conditions.
The A-8A’s front panel includes a rotary switch for input selection and potentiometers for adjusting the volume, bass, and treble. Interestingly, the bass control is boost only and the treble control, which includes a power switch, is cut only. The back panel has two RCA input jacks and screw terminals for connecting the load (4, 8, 12, or 16 ohms) and strapping the output transformer accordingly.
My A-8A lacks a bottom panel and, because the chassis includes rubber feet that would interfere with such a panel, it was probably intended for use without one. Unfortunately, in a modern high-EMI environment, this could be problematic for the higher-gain inputs.
On to the schematic. Being a rare amplifier, documentation for the A-8A is sparse. I was able to find a schematic, brochure, and parts list posted on Audiokarma at fairly low resolution. The “A” version of the A-8 includes an extra “preamplifier” stage, utilizing an extra 6SJ7, for extra gain. This stage is used only with the mic and magnetic phono inputs and has a separate RCA jack from the crystal phono input. Overall, its setup is unremarkable — it’s a simple common cathode pentode configuration with cathode bias.
Next are the input selector rotary switch and the tone control circuitry, which feed the grid of the second 6SJ7. It’s worth noting that the labels on this schematic are incorrect — the potentiometer labeled “treb” is actually the bass pot and the one labeled “bass” is the treble control. Although it’s not specified in the schematic or parts list, all three pots are logarithmic.
The second 6SJ7 amplification stage has a lower value plate load resistor than the first, but the preamplifier also drives the tone control circuitry and the volume pot, so the loads are actually similar. A 220 pf capacitor shunts the second stage’s load, presumably to reduce its HF response and ensure stability. Negative feedback is applied from the 16 ohm terminal of the output transformer to the cathode of the second 6SJ7, so the negative feedback is global when using the crystal phono input but not the mic and magnetic phono inputs.
There is an interesting trick employed here to maximize gain. When negative feedback is applied to the cathode of a tube, there must be some impedance between the cathode and ground. This is intuitive — if the cathode was fully AC grounded, there could be no AC voltage (and thus no signal) between them. Stages receiving negative feedback often omit a cathode bypass capacitor for this reason, but this entails a loss of gain to degeneration. The A-8A minimizes degeneration by employing two separate cathode bias resistors, the larger of which is bypassed with an electrolytic capacitor and the smaller of which is un-bypassed so that negative feedback can be applied.
The following two stages, which use both halves of the 6SN7, are a classic topology: common cathode triode directly coupled to a concertina phase inverter. In higher-power amplifiers, there is typically another differential amplification stage after the phase inverter. However, the low-power A-8A does not require this and instead uses the inverter to drive the output tubes. Notably, this configuration does expose the grid of the concertina triode to full un-loaded B+ voltage (about 450 volts) upon startup, since the directly-heated 5U4 rectifier comes up much more quickly than the other tubes.
The output stage is also fairly straightforward — the 6L6s are beam tetrode wired and biased with a 220 ohm shared cathode resistor. Bias is around -20 volts and the tubes are run close to the 20 watt plate dissipation rating of a standard 6L6G. For longer life, it may be desirable to use higher-wattage 6L6 variants (such as the 5881 or 6L6GC) which will operate more conservatively. I found the Peerless S516A output transformer in a scanned Peerless catalog, which specifies a 6600 ohm primary and 20 watts of power delivery. Frequency response is 30 hz to 15 khz -1 db, an unimpressive spec which suggests this transformer is a budget offering.
Based on the amplifier’s B+ voltage (measured with a multimeter), cathode resistor value, and primary impedance, the loadline should be approximately:
The first 5ish watts are class A. This loadline can deliver 30ish watts given enough drive and an appropriate transformer, but the A-8A tops out around 17 watts.
B+ is provided by a simple cap-input 5U4 supply; the output tube B+ does not even have a resistor for filtering. The power transformer is a Chicago PCC105 (see below for specs).
Based on this amplifier’s internals, I assume it was assembled by a competent hobbyist using a kit. Lead lengths and wire routing look too haphazard for a factory-assembled unit. Some leads were at risk of shorting and had to be re-bent.
The parts used are typical for the era — wax and electrolytic capacitors with carbon composition and wirewound enamel resistors. Wiring is point-to-point using terminal strips. Aside from one resistor discussed below, all parts appeared to be original.
The first step of this restoration was a recap; I replaced the wax capacitors and all the electrolytics except for the can capacitor. Because this amplifier is not worth a lot, I did not want to spend $45 on a new can capacitor, nor did I wish to undertake a time-consuming re-stuff. Each section of the can capacitor did exhibit somewhat high ESR, but not so high as to cause serious issues. Moreover, there was no leakage to speak of.
The 450 volt rating of the original Mallory can capacitor is a bit low considering B+ spikes to around 450 volts upon startup. But I assume Heathkit knew what they were doing; perhaps vintage multi-section can capacitors were known to tolerate startup voltage spikes above their rated value. If I were to replace this capacitor, I would use a 500 v part.
The potentiometers and switches needed attention, as each exhibited a varying degree of stickiness and excessive resistance. I applied liberal quantities of 3-in-1 penetrating lubricant to dissolve the dirty oil and then used a paper towel to wick the mixture up, repeating until each rotated freely. Because the rotary input selector switch is open-frame, its contacts were quite corroded, so I applied contact cleaner and scrubbed them with a paper towel.
Although the resistor between the input selection switch and the bass control circuit is specified as 100K ohms on the schematic, my A-8A had a non-original 1M ohm resistor installed in this position. I reverted this resistor back to 100K ohms.
In preliminary testing, the crystal phono input worked well, but the high-gain mic and mag phono inputs were being swamped by EMI. After filming the video, I continued working on the A-8A in an attempt to reduce noise and EMI pickup.
When I did the recap, I removed a .05 uf cap between mains and the chassis because it was not safety rated for line-to-ground operation. While troubleshooting EMI, I added a .01 uf X1/Y1 rated safety cap in the same position, but it did not seem to have any effect.
By shorting the signal to ground at various points in the amplifier and by observing the volume pot’s effect on EMI levels, I inferred that the noise was arising in the input circuitry (as opposed to resulting from insufficient power supply decoupling or h-k leakage in the first tube).
Luckily, this amplifier’s grounding scheme is relatively good. Although it doesn’t utilize true star grounding, nor does it have a single thick ground bus wire, circuit ground connects to the chassis at a single point — the RCA input jacks — and the power transformer HV secondary center tap connects directly to the filter capacitor.
The RCA jacks were dirty and corroded, which creates two issues: high resistance between the chassis and circuit ground, and high resistance between the shell of the RCA plug and the shell of the RCA socket. Because the amplifier is not mains earthed but my source (a DAC) is, this would also mean high resistance between the amplifier and mains earth. To solve the former issue without replacing the jacks outright, I added a spade to connect circuit ground and the chassis. I secured this spade with an RCA jack mounting bolt, as placing it farther away might have caused a ground loop. To reduce dirt and corrosion on the jack shells, I sprayed a new, tight-fitting RCA plug with contact cleaner and rotated it in the jack.
These measures did noticeably reduce EMI levels, allowing me to use the mic input without interference overpowering the input signal (see test results below).
While checking the power supply for issues that might lead to noise, I also discovered the large 4.7K ohm 2W decoupling resistor between the first and second B+ filter capacitors had gone up to around 7.1K ohms, so I replaced it with a modern 4.7K ohm 3W metal oxide part. This replacement had no effect on distortion, but I felt it was necessary given the original resistor had already begun to fail high and would probably continue to increase in resistance. Ideally, every carbon composition resistor in a critical position should be replaced with a modern metal film or metal oxide part, but sometimes you have to pick your battles.
I tried various tubes with this amplifier. In the end, I achieved the best performance with: NOS 6SJ7s, a new JJ 6SN7, and new factory-matched Tung Sol reproduction 5881s. Keep in mind the difference between the worst-performing tube combos and the best was only a few db of THD. I also placed a sheet of tin foil beneath the amplifier and clipped the foil to the RCA ground shell, simulating a bottom cover. All tests performed with an 8 ohm resistive load.
For starters, here’s frequency response @ 1 watt output. Ignore the Y axis on the graph — my REW/FlexASIO drivers corrupt themselves regularly, so I don’t bother going through the calibration process. Instead, I check output voltage using a scope.
Frequency response is good, albeit with some treble roll-off. There seems to be a resonance at 45-60khz.
Noise level (xtal phono input, no input signal):
The dBV figures for these spectrum analyzer measurements should be correctly calibrated. You can see there is substantial 60 hz and harmonics despite my noise mitigation attempts. Luckily, the hum doesn’t really intrude on the listening experience.
1 watt (xtal phono input):
Respectable performance at 1 watt, especially for a budget integrated amplifier.
5 watts (xtal phono input):
10 watts (xtal phono input):
15 watts (xtal phono input):
This is approaching the amplifier’s maximum power. Higher-order harmonics rise substantially.
18 watts, onset of clipping (xtal phono input):
At 18 watts, the amplifier begins to clip visibly on the oscilloscope. Maximum power before clipping is around 16-17 watts.
All the above screenshots were taken with signal applied to the xtal phono output. Here is a picture of 1 watt power output with the mic input:
1 watt (mic input):
Unsurprisingly, noise (and thus THD+N) suffer considerably from the very high gain and the lack of negative feedback around the preamp stage. However, distortion is barely worse than the xtal phono input.
10 watts (mic input):
Interestingly, distortion with the mic input is actually lower than the xtal phono input at 10 watts, albeit only slightly. To be honest, I’m not sure why this would occur. Noise remains substantially elevated, so the listening experience is still worse than the xtal phono input.
Lastly, square wave at 1 watt(ish), starting with 1 khz:
It’s not particularly square, but no sign of ringing.
Next, 10 khz:
Not great, but not bad enough for me to modify the amp. Let’s try capacitively loading it (.1 uf in parallel with the 8 ohm resistive load) to see whether it breaks into oscillation:
Little effect from the capacitive load besides a minor increase in ringing at the leading edge. The HF response isn’t too pretty, but the amplifier seems quite stable.
The A-8A is a fun little amplifier. It may not have the power, low distortion, or extended HF response of a high-end amp from the era, but it offers admirable performance given its diminutive size and relative simplicity. And, in subjective listening, it sounds quite good. The output is very slightly noisy, but the amplifier is remarkably warm and engaging thanks to the tube-y low damping factor, good low-frequency response, and slightly recessed treble. I hope to locate a Heathkit collector or a mono amplifier enthusiast so this A-8A can find a good home.