Musing about the HP Barney Oliver Amplifier

Status
Not open for further replies.

3v0

Coop Build Coordinator
Forum Supporter
Decades ago when there was still audiophiles I heard tell about an amazing solid state audio amp designed by HP engineers for personal use. At that time is was a old design. Today I ran into some info on it. According to the document below HP constructed these units. Not sure but some HP employee's may have built their units.
I don't think they were ever sold to the public.

Might be a fun build but I expect there are parts which are hard to get and the PCBs were possibly crated with chart pack tape on acetate.

http://www.hparchive.com/Manuals/Barney_Oliver_Amplifier_Manual.pdf

There is a complete unit on **broken link removed**for $1500. Too rich for my blood.

A note to my friends here: Good to see so many of you are still around. Although I still love electronics and uC's my health is better served by being more active.
 
Off topic a bit, but it's great to see you again 3v0. Been a while! Glad to hear you're doing well.
 
I had also had heard about The Legend of the HP Amplifier.

Now I've seen its schematic. I may now die in peace.
Thanks for sharing.

Interesting, the manual's format closely resembles the manuals for HP equipment of the era.
 
I imagine that an HP document writer got onboard.

One really neat thing about HP is that engineers were encouraged to work on unofficial projects.
 
i'm not sure that CR1 and CR2 are a really good idea. they are within the feedback loop, but putting nonlinear devices in the signal path is usually a bad idea. i'll try this in LTspice and see if it makes any difference. having two potentiometers for setting the offset seems kind of strange too...

EDIT: ok, tried that amp out in LTspice...

in case nobody is familiar with the use of two different rail voltages (40V and 50V), it's not uncommon in high end amplifiers and pro audio amplifiers.




1) had to add a capacitor in series with the input, or the offset pot on the noninverting input (Q1 base) doesn't do much.
2) adjusted with no signal to get about 4mV offset. NOTE: there's a DC blocking cap on the inverting input, will try testing without offset pots.
3) measured idle current, outputs running 60mA which seems a bit high, but not excessive.
4) tested just below clipping at 1khz, highest harmonic is 2nd (2khz) measured -87db from fundamental, which is about .004%
5) shorted the diodes between the output devices and emitter resistors, 2nd harmonic now at -91db which is 0.002%
if the diodes are to protect output devices from inductive kick, there's a better way to do it.

there are a few other oddities in this circuit. there's a lot of 1% tolerance resistors. considering this was a lab project, it's possible to have an abundance of 1% parts, that get used in places where a 5% part would be sufficient. usually in audio amplifiers, the only place they are used are in the feedback loop, where you want the voltage gain to match exactly in multiple channels.
second "oddity" is the current sources. they don't have anything as a reference, except a resistor. this makes for a "sloppy" current source.
third is the compensation caps, they go to ground rather than from B-C on the voltage amp.
4th, output stage biasing done by a huge string of diodes, and they're not thermally coupled to the output stage's heat sink.
5th, feedback taken from the outputs, and the center of the bias stack. usually connections from output to driver or predriver midpoints can cause catastrophic failures if one of the output devices shorts (as in "all the way back to the diff amp")
6th Q3 is "excess baggage", and it's sole purpose is as a current source for the offset controls.
7th the use of several RC networks in the feedback loop are there to alter the frequency response to resemble the response of tube amps. its better to do any equalization/tone control functions in a preamp.
8th the output of the voltage amp and it's current source seems to be a bit strange in the way it's done, and there's no capacitor between the two ends of the bias stack, so the drivers and outputs aren't exactly getting a symmetrical drive signal.
8th inductors in the diff amp????????


INITIAL TEST:

Av=52.77
Pout=71 watts into 8 ohm load
THD=.0045% @ full power
clipping not symmetrical, with + side clipping first, and 5 uS of recovery time.


edit: pretty good amp design for 1973. i just spent a couple of hours changing small stuff to see what would happen... negative half has a tendency to oscillate, which is visible as some ringing when it clips. that oscillation tends to crop up when any significant changes are made. stability is always the hardest part of designing amps.... you have to work real hard to keep it from becoming a transmitter...
 
Last edited:
I imagine that an HP document writer got onboard.

One really neat thing about HP is that engineers were encouraged to work on unofficial projects.
I talk with a engineer from energy micro (silicon labs), he was ex HP and he says that HP's tag line used to be 'we invent'. He was telling me that Energy Micro still has a policy that every friday you dress down, you go into work and as long as your using there products, you can do what you want. Apparently this was something they started when they first became a company as one of the head guys was also ex HP.

HP was some force years ago, looking at the history of printers they invented the ink jet and were years ahead of everyone else for ideas. Shame its a shadow of what it was.
 
HP at one point, was the world's foremost test equipment manufacturer.
With the exception of oscilloscopes, where Tektronix had the lead.
 
HP didn't invent everything they sold. Like many big companies, they bought out smaller companies and then put the HP label on the product. So, it's interesting to see that in the last several years the reverse is happening. The single HP behemoth has been split into HP (computer products), Agilent (medical equipment), Keysight (electronic test equipment), Avago (semiconductors). It's probably not a bad idea really since the product lines are vastly different and have different customer bases.
 
No but they did actively invent and people like 3V0 and others i know who worked there, will atest to the fact that 'playing' with stuff and following your nose was actively encouraged. The fact its now split over different groups and dosnt have the solid quality or reputation it did have is a real shame.

I have several different old HP logic analyzers with oscilloscopes etc on them, they are 20-30 years old and would still give many products a run for the money. Even Tex scopes have gone plastic Chinese cheap nobs. I have had to replace a power supply and HDD on my HP 16500c Logic analyzer, if you consider its age it had touch screen (colour) and no end of different cards you could slot in. A really serious bit of kit but the weak point is the HDD
 
I'm not disputing HP quality or their encouragement of their employees. But I think it would have been better for the test equipment division to have kept the HP moniker, and to have called the computer division Compaq. I think the name change did far more harm than any perceived decrease in product quality.
 
I agree totally, i used some old compaq pocket pc things for controllers before the r pi came about, they went down hill pretty quick though. I think it tries to be all things to all people now. Its a shadow of what it was.
 
Back on the subject of the Barney Oliver amp, in 1973 solid state amps were just starting to get a decent reputation, but they tended to be designed like op amps with an added power output stage, and suffered badly from TIM (transient intermodulation) distortion. The specs would talk about harmonic distortion, and maybe regular IM distortion, but not TIM distortion. I don't know if they even had proper test equipment at the time to measure it. This was eventually recognized a few years later, and the amp designs improved accordingly. I'd be interested in finding out how the Oliver amp fared in this regard.
 
HP at one point, was the world's foremost test equipment manufacturer.
With the exception of oscilloscopes, where Tektronix had the lead.
i got the chance to work with both an HP and a Tek scope side by side. Tektronix had rock solid, and predictable triggering, and the HP scope's triggering was kind of difficult to set. tektronics scopes used 1-2-5-10 steps in the gain and sweep selectors, HP used 1-3-5-10, which took a bit of getting used to, and i really didn't like that sequence.


The specs would talk about harmonic distortion, and maybe regular IM distortion, but not TIM distortion. I don't know if they even had proper test equipment at the time to measure it.

the root cause of TIM turned out to be slew rate limiting, and as amplifier design progressed, it became easier to deal with TIM by insuring the amplifier's slew rate was adequate. if an amplifier's power bandwidth was 50-100khz, the slew rate was high enough to eliminate TIM. that's why, in the late 70s, slew rate and/or power bandwidth specs began to appear in the documentation. for a 100W amplifier, a slew rate of 20V/microsecond or better is adequate to insure the amplifier isn't creating intermod products within the audio spectrum.

I'd be interested in finding out how the Oliver amp fared in this regard.
more testing to follow. i can say that running the design in LTSpice has shown up a slight instability on the negative side of the output waveform. since i don't have models for the original output devices, i'm using the "favorite" output devices from the 60s and 70s, the 2N3055/2N2955

slew rate seems to be about 40V/uS, so the amp would not have problems with TIM.
 
"HP didn't invent everything they sold. " Are they any big corporations that do ?

I think everyone agrees the test and measurement divisions would have benefited by keeping the HP name. The logic was that the computer people needed the HP name for marketing, more than we did. And I imagine they were right. We were allowed to use something like "Agilent, innovating the HP way".

Really it did not matter. Once Bill and Dave were both gone the HP way evaporated. We were competing with companies where engineers worked for peanuts. Hours got longer and deadlines shorter. The fun, and it was fun being a HP engineer, went out of it. It became another job.

Still it was a good run. You can't turn back the clock.
 
test and measurement divisions would have benefited by keeping the HP name
Agilent, innovating the HP way
Keysight
With every name change, and with every round of layoffs, and with every factory closed, and with every year that passes and no pay increase, and with ...... The HP way is going, going, gone.
In the last week I have driven past three plants that are closed and one that is almost there.
I need to keep my mouth closed. I have had pay checks from all three names. Probably no more.
 
i'm not sure that CR1 and CR2 are a really good idea. they are within the feedback loop, but putting nonlinear devices in the signal path is usually a bad idea.
if the diodes are to protect output devices from inductive kick, there's a better way to do it.
The diodes appear to help stabilize the output bias current while minimizing the output impedance and power loss in the emitter bias resistors.
The bias current sees 8.2Ω emitter resistors, but the large signal output to the speaker sees just the diode forward impedance.
The increase in distortion from the diodes is acceptable, since it's still very low due to the large negative feedback.
 
between having the diodes there, and not there doesn't seem to change the output impedance much at all (if 0.22R resistors are added in series to both diodes instead of a single resistor as shown in the schematic). the output impedance at 1khz is 13 milliohms without the diodes and 13.5 milliohms with the diodes. the diodes, even though they are inside the feedback loop add distortion. i realize this was a very early Lin topology design, and not many amplifiers of this type had been made. a lot of the resistor values are critical, which could mean that variations in transistor characteristics that occur in production would require some resistor values to be selected at final test.

there are a lot of amplifiers in production today that use a variation of the cascaded diff amp used in this design. however, the second diff amp in this design is somewhat of a "differential to single ended" converter, while most modern amps that use a second diff amp use both sides of the second diff amp to drive current mirrors and voltage amp stages.

this is a very "forward looking" design for 1971. most solid state amplifiers from the 1960s and going into the early 1970s had single-ended input stages, and quasi-complementary output stages.
 
Agree on the tight transistor characteristics requirement.
It was common for high-performance equipment transistorized designs back then, to have commercial transistors selected for certain parameters, and then apply "house numbers" to them.
Or at least a colored dot to facilitate parameter binning.

Heck, even my Zenith Transoceanic radio has selected transistors with Zenith house numbers.
 
Status
Not open for further replies.
Cookies are required to use this site. You must accept them to continue using the site. Learn more…