If anyone's ever been curious what ATI is putting into their amplifiers, here is a schematic I found. Generally speaking, their schematics do not seem to have been published. For those who aren't familiar with ATI, ATI is a fairly large US-based amplifier manufacturer that has not outsourced everything to China. They have OEMed products for a number of companies, although for the most part they don't admit whose products they make. Their calling card, though, is individual amplifier modules and you can usually tell their products from the back panel.
To be fair, this schematic is actually for the Earthquake Cinenova Grande circa 2014. They were kind enough to post it on their European website (by accident?). The amp is clearly an ATI. The original versions had the same goofy parallel port connector ATI introduced (and which went nowhere), the modules use the same heatsink as the ATI AT2505, the topology is a complementary diff which ATI is known to use, and so forth. Whether ATI still uses this exact circuit in their house branded products, I don't know. I suspect in lower priced OEM gear like the Monoprice Monolith they do (less a few output devices). Their new high end stuff is a fully balanced design, so is different.
This is a fairly simple design complementary differential input design, without a whole lot of tricks up its sleeve. It is what it is and perhaps someone will find it useful if they're trying to fix one. Or are curious how a particular design performs. (Bench tests at Product Review). I suppose they had to keep the parts count down to fit it all on the module? Who knows. There is a lot that could have been done here that was not. My guess is that this is close to the same thing (or even less sophisticated) than what Jim Bongiorno designed for SAE more than 40 years ago (which was owned by the same people as ATI).
To be fair, this schematic is actually for the Earthquake Cinenova Grande circa 2014. They were kind enough to post it on their European website (by accident?). The amp is clearly an ATI. The original versions had the same goofy parallel port connector ATI introduced (and which went nowhere), the modules use the same heatsink as the ATI AT2505, the topology is a complementary diff which ATI is known to use, and so forth. Whether ATI still uses this exact circuit in their house branded products, I don't know. I suspect in lower priced OEM gear like the Monoprice Monolith they do (less a few output devices). Their new high end stuff is a fully balanced design, so is different.
This is a fairly simple design complementary differential input design, without a whole lot of tricks up its sleeve. It is what it is and perhaps someone will find it useful if they're trying to fix one. Or are curious how a particular design performs. (Bench tests at Product Review). I suppose they had to keep the parts count down to fit it all on the module? Who knows. There is a lot that could have been done here that was not. My guess is that this is close to the same thing (or even less sophisticated) than what Jim Bongiorno designed for SAE more than 40 years ago (which was owned by the same people as ATI).
Attachments
Last edited:
Yup. Basically a leach or an Ampzilla with nary a clever thought put into it.
Still, it seems to be pretty much impossible to get a service manual or schematic for any of ATI's products, so hopefully this is useful to someone. I was interested in it because Jim Bongiorno had mentioned to me probably 15 years ago that ATI was still basically selling Ampzillas with all the complexity ripped out, and so they are. The parts are updated, but the circuit is almost as vanilla as it gets.
In all fairness to ATI, though, I have to think they improved the design in newer products that aren't OEM for Earthquake. The distortion is an order of magnitude lower. It would be interesting to dig up a schematic of that and see just what they did.
I took a close look at Ampzilla back in the early 1970s and sure it had a comp diff but
it sure was overly complicated with many electrolytic caps. Did not really like it, but if
there is something clever in there perhaps I should take another look.
it sure was overly complicated with many electrolytic caps. Did not really like it, but if
there is something clever in there perhaps I should take another look.
Ampzilla schematic:
Ampzilla Schematic
It has MIC, a Baker clamp, but the diff pair biasing is a mess with electrolytics.
Ampzilla Schematic
It has MIC, a Baker clamp, but the diff pair biasing is a mess with electrolytics.
JC seems to know the history and gives credit to Dan Meyer for at least first advertising it:
http://www.diyaudio.com/forums/lounge/200865-sound-quality-vs-measurements-561.html#post3027053
This is his Tiger .01 published in Mar 1973, note that it has diff pair degeneration:
http://www.swtpc.com/mholley/RadioElectronics/Mar1973/RE_Mar_1973_pg42.jpg
And the Tigersaurus from Dec. 1973:
http://www.swtpc.com/mholley/RadioElectronics/Mar1973/RE_Mar_1973_pg42.jpg
The large caps make the diff pair degeneration in the Ampzilla only effective for DC
bias, not for improved overload margin.
http://www.diyaudio.com/forums/lounge/200865-sound-quality-vs-measurements-561.html#post3027053
This is his Tiger .01 published in Mar 1973, note that it has diff pair degeneration:
http://www.swtpc.com/mholley/RadioElectronics/Mar1973/RE_Mar_1973_pg42.jpg
And the Tigersaurus from Dec. 1973:
http://www.swtpc.com/mholley/RadioElectronics/Mar1973/RE_Mar_1973_pg42.jpg
The large caps make the diff pair degeneration in the Ampzilla only effective for DC
bias, not for improved overload margin.
Hi all, I know this is an old thread. But it's been helpful. I have a couple schematic questions if anyone has a moment. Just want some confirmation on a couple things:
1) I assume V0+ & V0- are the same voltage as V+ & V- (somewhere around +/- 95-100vdc), although perhaps the input stage power voltages are filtered more stringently or even zener clamped? The input section needs the same high voltage headroom as the output since the backend is a "follower" downstream of Q16 & Q2.
2) Voltages OP-E & OP_C. I assume these voltages are more or less binary. Both near ground (or more) and the input section is up and running. If op_e is near -15v and op_c is near +15v, then the long tail current sources are off and the AMP is more or less standby. Any idea how they are used operationally in the amp?
3) I've simulated the circuit and found the bias spreader needed a bit of tweaking to make it go, but that may have to do with item (1) assumptions. The summation is I had to increase R142 by about 50% to get any sort of output stage bias current. I also have used generic BJT models and that may have skewed things bit too.
Thanks!
1) I assume V0+ & V0- are the same voltage as V+ & V- (somewhere around +/- 95-100vdc), although perhaps the input stage power voltages are filtered more stringently or even zener clamped? The input section needs the same high voltage headroom as the output since the backend is a "follower" downstream of Q16 & Q2.
2) Voltages OP-E & OP_C. I assume these voltages are more or less binary. Both near ground (or more) and the input section is up and running. If op_e is near -15v and op_c is near +15v, then the long tail current sources are off and the AMP is more or less standby. Any idea how they are used operationally in the amp?
3) I've simulated the circuit and found the bias spreader needed a bit of tweaking to make it go, but that may have to do with item (1) assumptions. The summation is I had to increase R142 by about 50% to get any sort of output stage bias current. I also have used generic BJT models and that may have skewed things bit too.
Thanks!