Class-A Push-Pull Tube Power Amplifier

ss-A Push-Pull Tube Power Amplifier

s Class-A Push-Pull Tube Power Amplifier uses a Pair of Push-Pull Class A, Ultra Linear Mono Block Tube Amplifiers that can be used with several different vacuum tubes including KT77 / 6L6GC / KT88 with a 12SL7 driver and 6NO30 tub
plifier stage is based on the Compact Hi-Fi Power Amplifier. One thing about DIY audio is that it is a journey, not a destination, it never ends. One project leads to another. The only limits are time and money. DIY audio is a lot about per
le I was quite happy with my previous tube amplifier projects, I felt there was room to improve (here comes the journey again). I like to be involved in the music. If anything sticks out, it will degrade the experience. So I tend to like s
ponse, lots of detail, wide soundstage and full spectrum of sound. These amps deliver all that in quantity. Regardless of what tubes I used for outputs, the sound is "silky" and refined.
he previous project, the OddWatt 225 (6SL7 SRPP / KT77 Class-A Push-Pull Tube Amplifier) is in daily operation and (with JJ KT77
bes) has a wonderful tube sound. The Odd Blocks are rather similar only they are two separate amps and were planned up front to
ble to use any of the power tubes that share the same pin arrangements (6L6GC, KT77, KT88, EL34, 6CA7 and all the variants).
ubeDepot.com has a good selection of these tubes - NOS and new productions. I initially used KT77 tubes as I was familiar with the
und and it would be a good starting point. I tried the amp with Electro-Harmonix KT88 tubes. Using the EH KT88 tubes has been a
ixed situation. The EH KT88 can deliver more power, but the sound is different from the KT77. Additionally they don't behave well in
is particular amp. I suspect the reason is related to thermal problems and the use of common (joint) cathode bias. The circuit allow
r adjustment of the bias (the 25 ohm variable resistor) but the EH tubes do not stay in sync. One or the other tube (not predictable
hich one) will start to conduct more than its fair share of the current. With current balance being an important consideration of the
esign this is not good. By comparison the JJ 6L6GC tubes and JJ KT77 will balance within one mA and stay there. It is possible that t
a situation with just the one brand of KT88s. I have since then acquired a set of new production Gold Lion KT88 tubes and they beh
erfectly. I will not further speculate why the one brand works and the other doesn't. It is just the way things are. For a more comple
escription of how the two stages of the amp work see the original OddWatt project (ECC802S SRPP / EL84 (6BQ5) Push-Pull Tube
mplifier). The SRPP driver and SIPP output work very well together. An alternative driver for 9 pin application is the type 5751. This
be will directly substitute for the octal 12SL7 in this circuit. Other tubes with the same base as the 5751 (12AU7, 12AT7 and 12AX7
ere tried, but none performed as well as the 5751.

NO30 Thermal Delay Relay Vacuum Tube

hotograph 2: 6NO30 Thermal Delay Relay Tube

s in nearly all tube projects dangerous voltages are used. Contact with the voltages at various points in this circuit can be fatal. If yo
e not familiar with tube equipment or are not experienced with high voltages, perhaps this is not a suitable project.

his is a pair of mono blocks amplifiers - each with its own power supply. In some ways this is more complex and costly and in some
ays cheaper, easier and has advantages. For starters if you only need one amp, then you need only build one. Second if you need th
erhaps one for a sub woofer) then you can build three. In the stereo version it requires two power transformers, rectifiers and filter
ut the transformers are less expensive than the one needed for a single chassis stereo amplifier. A side benefit is that you don't nee
t of wheels to move the amps. A single chassis amplifier would weigh in excess of 50 pounds.
RPP / SIPP Tube Amplifier Design with CCS

or design purposes I found that potential power tubes pretty much fell into two groups. Group "A" consists of KT77, 6CA7, EL34, 6L6
nd similar variants. Group "B" consists of KT88, KT90, 6550 and their variants. The difference is in both the bias voltages and maxim
ower dissipations. In the measurements and listening sections I relate the differences. The basic circuit is the same as previous
ddwatt amplifiers (see other projects). I saw no reason to mess up a good thing. There are some refinements though. The CCS in th
iginal amps was a standard LM317 voltage regulator IC. I have since switched to the LM317HV. The voltage breakdown margin is
etter. I understand that these are discontinued but there are many available in parts houses. It really is only needed for the "B"
ersion. The "A" version can use the standard LM317. To date I have not been informed of any failures in either application, but I don
el comfortable with the 37 volt rating of the standard parts in the KT88 application. Some resistor values in the SRPP stage have be
hanged to increase in the current flow. Other DIYers have determined that this reduces the distortion level a bit. It makes sense, bu
easurements to date have not confirmed it as the level was already quite low. The power supply uses a Edcor 180-0-180 transforme
th a full wave bridge across the full winding. With capacitor input filtering the B+ is between 450V and 475V. I tried a one Henry ch
r a filter and decided it wasn't worth the cost and failed to significantly improve the filtering.

2SL7 SRPP KT88 Push-Pull Tube Amp Schematic

gure 1: OddBlocks - 12SL7 SRPP / KT88 Push-Pull Tube Amp Schematic

2SL7 / KT88 Tube Amp Power Supply Schematic

gure 2: OddBlocks - Power Supply Schematic

art of DIY is that you can do things that manufacturers won't. I found a source for 6NO30 "tubes". For those of you who don't recogn
this is a thermal relay in an octal socket. This is a rather old technology gizmo. It is the same size and shape as the 12SL7 and thu
dds a nice symmetry to the amps. These "tubes" work on the principle of a heating element and bi-metal switch. The contacts are ra
r 500,000 operations at full power. Unfortunately the voltage rating of the contacts is 120 VAC (breakdown voltage rating is 500 vo
hen new - whatever that means). Clearly, this is not high enough for the 360 volts AC being switched. So I use the contacts to activ
12 volt relay that actually does the work of connecting the AC (360 v) to the rectifiers. The 6NO30 delays the activation by 30 secon
ther values of thermal relays are available as well. These were cheap ($2.96 each) NOS. The numbering for them is as follows: The
git(s) is the heater voltage. Then follows either NO or NC, for Normally Open or Normally Closed for the internal contacts. Finally th
st numbers indicate the seconds of delay. Using the thermal relay is probably a sign of madness and the job could be done simpler
transistor and a few passive parts, but it sure is cool. The remainder of the PS is similar to my other projects. I use a generic,
expensive 12 volt, 3 amp SMPS to power heaters as it is cheap, very compact and provides clean well regulated power. A conventio
ansformer, rectifier, and filter could be used. The windings on the Edcor power transformer are 6 volts and would be satisfactory for
oup "A" tubes. For group "B" tubes as bigger transformer is needed. Generally, it is a bit easier (and cheaper) to get the required
ower at 12 volts at 2 amps than at 6 volts and 4 amps. Losses are lower for one. AC could be used on the heaters as well. It would
ost likely increase the noise floor a bit. If you are at all like me, you hate hum and noise. I would stick to SMPS for the heaters in al
ut "budget" builds of these amplifiers.
onstruction - OddBlocks Tube Amplifier

he build is pretty straight forward. The chassis is steel by Hammond purchased from Antique Electronic Supply. The transformers ca
om Edcor. Wiring was point to point and I used some terminal (aka turret boards) boards for connections. The more sensitive circuit
placed as far from the power transformer and noise sources as possible. The filter capacitors and B+ relay were glued to the chassi
gnificant feature is the use of a ground buss (seen as a large bare copper wire in the photos). This has worked well in the past proje
r minimizing hum, noise and the possibility of ground loops. A special precaution is required with regard to the heater circuits. I
scovered some time ago that it is easy to exceed the heater to cathode voltage ratings of the driver tube when you use a SRPP desi
here are several possible solutions. The one I used is to bias the heaters at about one third of the 240 volt supply. There is no curre
ow (just a tiny bit of leakage - in the microampere range). There are two extremely important considerations with this method. One
art of the heater circuit can be grounded. Second, the source of the heater power must also not be ground referenced. It needs to fl
rt of like the heater circuit for many tube rectifiers. If you use a SMPS it needs to have the output isolated from the input and not in
ny way referenced to the ground. If it is not, then failure is virtually guaranteed. Another important point in the construction is the
nnections to the CCS. The "adjust" terminal of the LM317HV is grounded. The sense resistor is between the output terminal and
ound. This is different from the use of the IC as a voltage regulator. When properly connected, the current is equal to 1.25 divided
e value of the resistor in ohms. Thus 10 ohms will result in a 0.125 amp current flow. The CCS needs to be heat sinked as it has to
andle up to 8 watts of dissipation (when using KT88 at 180ma - less for other tubes and currents). Now for some magic, if you use a
xed 10 ohm resistor in the CCS for a basic setting it will result in close to 62 mA per tube. If you have a 22 ohm resistor connected i
arallel to it with a simple SPST switch you can bump up the current to 90 mA per tube. The first setting is perfect for group "A" tube
nd the second for group "B". Tube swapping is very easy in this amplifier.

2SL7 SRPP / KT88 Push-Pull DIY Tube Amplifier

hotograph 3: 12SL7 SRPP / KT88 Push-Pull Tube Amplifier - Inside View

or additional tube amplifier design and construction tips, see my design and construction tips and suggestions for vacuum tube
mplifiers. Also, I have posted some suggestions for a tube amplifier wiring color code that you can use during construction.
esting and Operation

efore actual listening to the amp it is important to verify that heater and B+ supplies are operational. There are lethal voltages in th
mp, use extreme care when making any measurements. I suggest that you install the 6NO30 (if you use one) and not any of the oth
bes. Turn on the power and check the voltages. There should be 12 VDC between the heater pins on the 12SL7 socket and B+ arou
75 at the filter capacitors. The 240 voltage point will be high (close to the 475 reading) as there is no load on the circuitry. There sh
e no B+ until the 6NO30 switches (about 30 seconds). If all is well, shut the power off. The filters will be fully charged at this point s
e careful not to contact any part of the circuitry. Insert the remaining tubes and reapply the power. Watch for problems (tube plates
owing red, sparks, smoke, noises and other fun things that indicate bad news). Check the voltages again. They should be in the pro
nges (12 VDC on the heaters, B+ of 475 and 240). If they are greatly different, something is wired wrong. If all is OK, shut down th
ower and attach speakers. Turn on the power. There should be little or no sound (hum or noise) coming from the speakers. A small
mount may be OK if you have very sensitive speakers. If you can hear it at 12 inches from the speaker, then there is probably a gro
shielding type of problem. I have 93 dB/watt speakers and the hum and noise is barely audible with your ear directly on the speake
ill. If this all checks out, then feed the amps a signal and see what comes out. It should be warm sweet tube music. Now is the time
o an initial balance on the output tubes. A volt meter across the test points should read in the 40-200 mV range. The object is to ad
e control to make the reading for each tube the same. The readings should add up to a number equal to the mA that the CCS is set
r. If you selected 125 mA, then the sum of the two readings should be close to 125. It is not important to be exact at this time. Allo
e amp to run for at least 15 minutes and check the setting again. It will probably have changed a little - so reset it. I suggest that y
heck it this way about once a month to account for tube ageing. If the tubes won't balance then swap them. Some tubes are more
nsitive to having the grid on the ground than others. It is nearly always possible to get them to balance. If the tubes won't balance
en there is probably either a wiring error or a tube that is out of tolerance. It is not necessary to have matched tubes, but if the pri
ght then it makes them closer and easier to balance. I have found that not all companies balance tubes the same way and sometime
matched" sets don't work as well as randomly selected ones. This circuit balances them at a fairly high level of idle current. Many
mpanies use a small current value or base the tests on maximum voltage and current. For what it is worth, I have had good luck w
tubes and not so good experience with EH KT88 tubes. Just to be sure no one gets too excited, it may not be EH's fault. Not all tub
ork well in all circuits and the EH KT88 tubes may be just fine in other designs. The ones I have just don't work well in this instance
he Gold Lion KT88 tubes worked fine, so it is not a basic property of the KT88 tube design itself.

ne area that always seems to cause problems is the phase of the feedback loop. In one orientation it will be "negative" and in the ot
will be "positive". Negative feedback stabilizes the gain and response of the amplifier. Only a small amount is used. It also takes ca
any possibility that something will excite the output transformer resonances at around 60 kHz. Positive feedback will do just the
pposite. The easiest way to tell without test equipment is to listen. If the feedback is negative the output will be a bit lower than if it
sconnected. If the output is louder is positive. This is easy to see on a scope when you feed in a small signal. With the small amoun
sed it will be just slightly lower in volume (about 3-6 dB). I received word from another DIYer who used different output transformer
s high frequency response was down quite a bit. In this instance, I suggest trying lower values of the NFB capacitor (1000 pF). The
xact values for the resistor and capacitor are based on the Edcor transformers. If you use some other brand you may have to
xperiment with the values.

2SL7 SRPP / KT88 Push-Pull DIY Tube Amplifier

hotograph 4: Odd Block Tube Amp with Cover Removed
easurements - Push-Pull KT88 Tube Amplifier (12SL7 Driver)

easurements were made using a HP 331A distortion analyzer, Tenma Low Distortion Signal Generator (0.05% residual THD)and a
elleman PCSU1000 (PC interface) oscilloscope. The results from testing the amplifiers are quite similar to the results from the OddW
25 (6SL7 SRPP / KT77 Class-A Push-Pull Tube Amplifier). That is not really a surprise as they are rather similar. With an 8 ohm resis
ad the response measured from 8 Hz (the lowest I can generate) to 19 kHz (at -1 dB). The minus 3 dB point was 30 kHz. The Gold
T88 tubes were up +0.5 dB at 10 Hz while the JJ KT77 tubes were flat at that frequency. Distortion was 0.14% or less at one Watt (
ter comment) throughout the whole frequency range. For group "A" tubes output before distortion was seen on the oscilloscope was
atts depending on the specific tubes used. Up to 20 Watts were possible. With group "B" tubes the corresponding output was 24 Wa
nd around 30 with noticeable distortion. As expected the distortion was virtually all harmonics. Typically they were suppressed by
ound 50 dB. Hum and noise were below -74 dBV which is the noise floor of my present test set up. It was virtually all wide band no
one Watt, the distortion harmonics were buried in the noise floor. For those of you who are graphically inclined, I included a spectr
nalyzer shot of a 1000 Hz sine wave and a few square wave images. I only included square wave response images as the sine wave
nes were perfect copies of the input signal at all frequencies. When I wanted to run a full distortion profile I ran into a problem. Just
efore I started I discovered noise on the AC mains in my shop. The noise tended to confuse the HP analyzer. The initial results were
elow the 0.15 % range and until I can clear up the noise problem I don't want to make any additional claims. In its place I ran a bun
high power scope tests. The KT88 version was able to deliver 24 watts of square waves from 20 Hz to past 10 kHz. They looked a l
ke the ones at one Watt.

his is as I have noted on several occasions a difficult to address topic. What I hear and like is almost certainly different from what
meone else hears and likes. I like to be involved in the music. If anything sticks out, it will degrade the experience. So I tend to like
mooth response, lots of detail, wide soundstage and full spectrum of sound. These amps deliver all that in quantity. Regardless of wh
bes I used for outputs, the sound is "silky" and refined. At the same time it has a very solid bottom end. This is nearly good enough
e to turn off the subs. Yes I admit to using subs. Two huge 7 cubic footers with 15 inch drivers. They are powered by a DIY LM3875
on-inverting chip amplifier (gainclone) kit and the signal is separated by a 24 dB/octave Marchand electronic crossover at 50 Hz. The
ain speakers are a pair of modified Altec Lansing 1010. The sound using just the mains (full range with no crossover to subs) is full
lid. There is plenty of bass for anyone but someone like me. I love those bottom two octaves. Imaging is excellent, very stable
ntering, separation is limited only by the other equipment (they are mono blocks after all), detail is excellent, and the sound has th
uality of warmth that you would expect from a tube amp. My personal feeling is that the JJ KT77 tubes have the most refined sound
hile the JJ 6L6GC and the Gold Lion KT88 tubes have the best bass. The KT88 tubes are of course louder.
ter Thoughts

always have ideas after the project is done and working. For instance is it easy to build? As a DIY project with point to point wiring,
ot particularly easy for a novice. For a more advanced builder, it would be easy. This is not the best choice for a first project. Is it wo
e cost and trouble? I think so. The transformers and tubes are the main expenses. They will run about $150US for each channel. Th
maining parts are up to the builder. It can be as little as $75US or as much as you want to pay. I used mid to upper grade parts an
timate it was about $150US per channel for the parts. Premium tubes (Gold Lion KT88 etc) will cost more. Can you buy this quality
und in a tube amp for $300 per channel? I think not. I will not get sucked into if SS amps will do the same for less cost argument.
for other people to decide on their own.
ated Links

Accurate LC M
ESR Meter / Transistor Tester / LC M
BA1404 HI-FI Stereo FM Transmi
500mW FM / VHF Transmitter Amplifier / Boo
USB IO Bo
60MHz Frequency Meter / Coun
1Hz - 2MHz Function Gener
Voltmeter Amm
USB Voltm
Accurate 0-500MHz RF Power M
Audiophile OPA2132 Headphone Amplifier
DS18S20 Dual Temperature M
BH1417 Stereo PLL FM Transmi
50mW BH1417 Stereo PLL FM Transmi
Phone FM Transmi
TV Transmitter with Au
5 Watt FM Ampl
TDA7000 FM Receiver / TV Tuner / Aircraft Rece
NJM2035 HI-FI Stereo Encoder / Multiplexer