cello OTA-1

GroupDIY Audio Forum

Help Support GroupDIY Audio Forum:

This site may earn a commission from merchant affiliate links, including eBay, Amazon, and others.

mikep

Well-known member
Joined
Feb 18, 2006
Messages
450
Location
Philadelphia
I hope this is ok to post, I think this company is out of the gear market at this point.

EDIT: I took the schematic down, I feel weird broadcasting it to the world. If you want to see it PM me.

this is the simpler of two discrete op amps used in the pallette EQ.

cello%20ota-1.JPG
 
..or perhaps the legending is correct, but just how it's drawn... -does the arrow on the emitter of Q4 need to go the other way round... and does Q5 need to have the arrow reversed AND the Collector/emitter swapped...

Wait.. I'm very confused!

Keith
 
woops. I had the emitter and collectors swapped on the output pair. suprised the sim worked at all . now it works with the original compensation values. I updated the original schematic. sorry about that!

EDIT:
I'm pretty sure that was the only thing wrong. I agree that it still doesn't look right, but this is why I posted it! the equipment this came out of certainly works. sounds amazing too. any comments?
 
Kinda cute. He (I believe it was either Dick Burwen or Tom Colangelo?) is using signals from both drain and source of the FET, something I did a while back in a second stage of a complicated charge preamp for UCLA.

The PNP below the FET drives Q1, a common-base stage, which drives the load of Q3/R3. Q3 is an emitter-follower biased with the current of J505, and drives Q5/R6, comprising an inverting transconductance stage.

Meanwhile upstairs J1's drain load, R1, drives another inverting transconductance amp of Q2 and R4, with a operating point mostly determined by the J505 current. Q4 loads the Q2 collector/current reg diode node and is another inverting transconductance amp.

The loop gain will be very load-dependent, since there is no lowish-Z output buffer that is typical for opamps.

The total number of inversions through each path to the output node is odd, hence the overall gain is inverting from gate to output. There is no convenient non-inverting input.

What is the phase/gain etc. performance from your sims?
 
[quote author="bcarso"]
The loop gain will be very load-dependent, since there is no lowish-Z output buffer that is typical for opamps.
(snip)
What is the phase/gain etc. performance from your sims?[/quote]

it seems to not care too much about load Z. it will drive 500 ohms about the same as 100k. there is about 50mA standing current in the OPs. I didnt measure the actuall current but these things do get pretty hot.

acording to the sim, open loop gain is about 80dB. open loop FR is -3dB at about 7kHz. it is unity gain stable. unity gain crossover point is at something like 40MHz

I bet this thing would make a good virtual earth mix amp.
 
[quote author="bcarso"]The loop gain will be very load-dependent, since there is no lowish-Z output buffer that is typical for opamps.[/quote]
But is the thing itself willing to be an 'opamp' ?

With the Wikipedia-opamp-definition-thing in mind from the recent thread NYD started about discrete opamps we may hit or miss the correct wording here and get a bit confused instead of getting things more clear, but after all thing above calls itself an OTA.
And I'd say an OTA is usually a VCCS (or CCCS), so high output-Z.

http://en.wikipedia.org/wiki/Operational_transconductance_amplifier
Hmm, differential input they say there...
....shouldn't always be required to be called an OTA.

Bye,

Peter
 
[quote author="mediatechnology"]Peter: I don't think Cello really meant OTA in the sense we know it. Maybe just a part number. This came up in another thread recently[/quote]
Hi Wayne,

OK, I see. Missed that one. It looks indeed like OTA is used for more than one thing, which is something I also recognize at work - some people have the same habit as Mr. Cello has.

Perhaps, with all due respect to mikep and Cello we should call it an "oddball transconductance amplifier."
:grin:
:thumb:

Let's find a few more OTA's :guinness:
Hmm, how about a Vactrol-based resistive divider being a Optically Triggered Attenuator ?
It won't be hard to find better ones...


Bye,

Peter
 
Is it an opamp? Well, in the original meaning of the term, as amplifiers with a lot of inverting gain and a single input, Yes---differential inputs were not the norm for use in analog computers. See some of the very early Philbrick stuff for example, including their tube opamps.

Is it an Operational Transconductance Amplifier? Yes, within a sort of broadminded intent: Voltage in, Current out. The output impedance, sans load and feedback network, is relatively high for the amount of current coursing through or available at least. A voltage-to-voltage transfer function is obtained when you terminate with an impedance.

But when RCA started to tout the OTA---in the form of their CA3080 for example---which has differential inputs as well, they also emphasized that it could have its transconductance set externally via a programming current pin. Thus this feature became associated with many folks' notion of a transconductance amplifier. And you could do many more nonlinear processing tricks, like variable gain, given this feature. Clearly there is no such control pin on the Cello module.

In that sense, and in connection with NewYorkDave's simple opamp thread, a single common-emitter transistor is practically a transconductance amplifier. But it has a highly nonlinear and relatively low input impedance, which was one of the significant departures to deal with for engineers who had learned to use vacuum tubes. The single Q can also be thought of as a current amplifier, Ib in and Ic out.

On cost-strapped systems with modest performance needs I've used a single transistor as the only gain element in a given circuit section---in one case I got a full-wave rectification function with two outputs in opposed polarity for use with a compressor. I count that circuit as one of the most satisfying things I have ever done. Not only did it work well, but it was wonderfully obscure-looking on the schematic.
 
[quote author="bcarso"]---in one case I got a full-wave rectification function with two outputs in opposed polarity for use with a compressor. I count that circuit as one of the most satisfying things I have ever done. Not only did it work well, but it was wonderfully obscure-looking on the schematic.[/quote]
Hey Brad, it's you who has designed the MXR Dynacomp ! :thumb:

But serious & FWIW, your description made me think of
these circuits : Q2,3 following Q5.
Some BOSS FX-pedals (SG-1 Slow Gear, NF-1 Noise Gate) do an alike thing.
Dunno your circuit looked more obscure though.

Regards,

Peter
 
[quote author="bcarso"]Is it an opamp? Well, in the original meaning of the term, as amplifiers with a lot of inverting gain and a single input, Yes---differential inputs were not the norm for use in analog computers. See some of the very early Philbrick stuff for example, including their tube opamps.

Is it an Operational Transconductance Amplifier? Yes, within a sort of broadminded intent: Voltage in, Current out. The output impedance, sans load and feedback network, is relatively high for the amount of current coursing through or available at least. A voltage-to-voltage transfer function is obtained when you terminate with an impedance.

But when RCA started to tout the OTA---in the form of their CA3080 for example---which has differential inputs as well, they also emphasized that it could have its transconductance set externally via a programming current pin. Thus this feature became associated with many folks' notion of a transconductance amplifier. And you could do many more nonlinear processing tricks, like variable gain, given this feature. Clearly there is no such control pin on the Cello module.

In that sense, and in connection with NewYorkDave's simple opamp thread, a single common-emitter transistor is practically a transconductance amplifier. But it has a highly nonlinear and relatively low input impedance, which was one of the significant departures to deal with for engineers who had learned to use vacuum tubes. The single Q can also be thought of as a current amplifier, Ib in and Ic out.

On cost-strapped systems with modest performance needs I've used a single transistor as the only gain element in a given circuit section---in one case I got a full-wave rectification function with two outputs in opposed polarity for use with a compressor. I count that circuit as one of the most satisfying things I have ever done. Not only did it work well, but it was wonderfully obscure-looking on the schematic.[/quote]

Agreed I ASSumed OTA name was just a coincidence or marketing distortion since most OTAs I've ever encountered were configured as multipliers (i.e. input diff voltage x bias current) to effect sundry voltage or current controlled circuits.

Further, common emitter output stages are notorious for being difficult to stabilize in power applications thus their near total absence from power amp designs despite their benefit of pulling within a saturation voltage of the rails. Years ago I made a crude low power amp (for my sister) using common emitter outputs driven by an opamp’s PS draw (design approach from some electronics rag). It was far from Hi-fi due to needing way too much compensation, but marginally better than the circa ‘60s consumer amp it replaced in the former's gutted chassis.

If we want to reduce this simplest opamp to absurdity a basic CMOS gate can act like an inverting amplifier, but is little more than a bench curiosity.

While there is an elegance to simplicity, the complexity of modern integrated circuits is not without purpose or benefit. IMO it's a little like comparing an old model T to a modern automobile. Sure you could increase the displacement of a model T to make as fast as a modern car, but without the refinements that increase the complexity we have become accustomed to that new improved T probably wouldn't make it around the first turn, and don't even try to hand crank 400 ci. (or was that a model A? I'm not that old).

JR

PS: BTW a single common emitter transistor may be more like half a Norton Amp (ala LM3900). Add a two transistor current mirror and you could have your differential input. I like to think of Norton amps as the back 2/3rds of a proper opamp.

One could hang a nice pnp or p-channel differential in front of a Norton amp to make a hybrid opamp. The LM3900 didn't deserve that much effort but I recall a higher performance LM359 that appeared worthy. No obvious choices for that task, and you couldn't call it discrete if that's important..
 
[quote author="bcarso"]Is it an opamp? Well, in the original meaning of the term, as amplifiers with a lot of inverting gain and a single input, Yes[/quote]

thanks for the circuit analysis, brad. makes a bit more sense if you consider the signal coming out of the fet source. By the way, I guessed at grounding the base of Q1. that comes out to a pin and I don't know what the original implementation was doing with it. can you think of any use for that node, other than just grounding it? maybe a good place to inject a servo voltage?

why is this configuration so (seemingly) rare today? IMO this could be more useful than a current feedback non-inverting amplifier. are there textbook examples of how to construct a hi gain inverting stage that Ive never noticed?

There was a post in the recent (NYD's) simple opamp thread talking about 3 basic types of amp. the cello equipment this amp came out of has a total of 3 building blocks. this one, a unity gain voltage follower and a full blown opamp with a differential input (cascode FET). so that is cello's idea of the basic 3.

[quote author="John Roberts"]
Further, common emitter output stages are notorious for being difficult to stabilize in power applications thus their near total absence from power amp designs despite their benefit of pulling within a saturation voltage of the rails. [/quote]

aha! I bet the output swing capability was why they chose this. the equipment this came out of has big time headroom. it is gain staged to have lots of internal headroom (presumably so you can boost with the EQ and turn down the output instead of the input. idiot-proof) and it can put out a little over +30dBu.

mike
 
[quote author="mediatechnology"]mikep: What was it about the Cello that people liked? I don't know that much about them. Do you suppose the "basic three" blocks contibuted to its' signature?[/quote]

Ive never talked to anyone else that has used cello gear professionally so I don't know what other people think of it at all! Im actually lucky enough to be the primary user of this EQ so I can comment on the subjective quality. It is in my mastering studio. I do not own the studio or the EQ, Im just the engineer. Ive been using it for about 2 years and just recently had to pull it to repair the PSU.

the signature is very clean with the slightest bit of airyness perhaps. not a "sheen" per say. If i had to say which side of clean it strays from, it would be bright rather than warm. but it is very subtle. you can take off 1 dB at 26k and then it sounds slightly warm. it has an effortless quality to it, especially when boosting lows. incredible headroom. I often use it as a line amp right before the A/D because it has independant L and R 1/2 dB trims and I have come to believe that it can do no harm to my masters.

the time I put it on the analyser I could not measure ANY harmonic distortion at any setting. my set is good to about .002% also listening to the residual its just pure hiss.
 
I don't remember all the details of the Cello story, but you may recall that it was founded by Mark Levinson after he allowed Madrigal Audio to deny him any further use of his name as a brand (!). It's interesting (well, to me anyway) that Sidney Harman finally persuaded Sandy Berlin to sell Madrigal to Harman. The company was even kept more-or-less intact for a while. Now it's more important to Harman, economically, as a brand (that is, Mark Levinson) for use in Lexus vehicles. They had a massive exodus when they began to lose a little money and the QA and bean counter types descended, and although they still make products I don't know if anyone is left from the company as it was when acquired. My 'phile friend Don thinks they began to go to hell in a handbasket when they started using IC opamps some time ago.

My recollection about Cello was that Richard Burwen had a lot of input and he believed that parametric EQ had its place, but was almost always low-fi and often lacked good control for creating really gentle slopes that he felt were often needed. The Cello Pallette was introduced to address these issues, and I recall the sort of hue-and-cry about TONE CONTROLS (said with mock horror) when the unit was reviewed in periodicals like The Abso!ute Sound---followed generally by an endorsement based on what the reviewer was persuaded to hear when he used the device to change things a bit, usually correcting for problems in the source material.

As far as whether there are any specific advantages to the OTA topology, I don't know. Clearly it works; the stability issues alluded to by JohnRoberts are tractable partly due to the substantial emitter resistors (his point being well-taken with regard to power amps though, a complex subject in its own right). The parts used are no longer state-of-the-art but the performance enhancements from such would likely be fairly minor.

mikep: the idea of that base lead being a servo or at least an external bias adjustment is quite plausible I think. JFETs are notoriously all over the map in pinchoff voltage and saturation drain current, so if low offset was required an adjustment or servo would be in order.

There are so many ways to build gain stages that it would fill many books to describe them all. It's almost astonishing that new variants continue to be discovered and even to have utility. Many of them are re-discoveries because no one could have read all of the literature, and as well plenty of designs have been kept proprietary, rather than published or patented. Also so many circuits were thought up before the devices existed to make them practical, so unless you go way back you will miss them. It's disconcerting when you find out that your idea was disclosed in the late 1930's, or it's somewhere in the Langford-Smith 4th edition.
 
[quote author="mikep"]I hope this is ok to post, I think this company is out of the gear market at this point.

EDIT: I took the schematic down, I feel weird broadcasting it to the world. If you want to see it PM me.

this is the simpler of two discrete op amps used in the pallette EQ.

[/quote]

Is Cello not way long gone? Seems like Mark Levinson at least has been through a couple of companies since (Red Rose I believe was one). If there's still someone out there who feels proprietary about those designs I'd be rather surprised. But then see that thread about the automatic renewal of copyrights.

I didn't capture the image, but I stared at it enough that I've memorized it. BTW, the 3819 is not a very good approx for the FET you mentioned, which is a "process 51" part with appreciably higher transconductance and input C. EDIT: oops I may be conflating this with another thread regarding the FET---is it really a 3819? Also I lied---I find that I did save it.
 
[quote author="clintrubber"]
Dunno your circuit looked more obscure though.

Regards,

Peter[/quote]

Well, again, this was ONE npn Q and two diodes and R's and C's which produced simultaneous positive and negative-going full-wave rectified voltages (at a fairly high impedance), to drive a four-Q current steering cell, which was part of an underdamped sort-of-Sallen-Key highpass. It is inside a speaker called Duet which I think was branded JBL. Truly a "gutless wonder", as my late father used to say, referring to some of his own stuff, approvingly.
 
I don't know Brad, I just don't want to be "that guy". Ill share the schematic with anyone else who wants to try to learn from it (like me), or who wants to try it in their own noncomercial DIY projects.

Tom C. is still making equipment
http://www.violalabs.com/index.html

and this?
http://www.cello-audio.net/intro.cfm
check out the "reflections" section on this site. lots of pictures of cello gear.


there is also some people (in washing state I believe) doing system design/integration under the Cello name. I think they had something to do with the original company as well but I'm not sure.
 

Latest posts

Back
Top