Oriel Z1, my headphone amp, or...

[Electrovoicer]

... how to try to make a business and finally spend a lot of money and get a good fun, no business
I want to share my experience.

The idea was to develop endless quality amp for private purpose.
I've went the long way trying something.
My end point for this moment is amplifier based on the following rules:
- semiconductor-based amp with negative voltage feedback
- constant open loop gain in audio range
- A-class output stage
- short and exponentially decreased harmonics spectrum
- differential input
- separated smps's for l/r channels
- ascetic but modern user interface

This amp has the following features controlled with menu:
- Differential or single-ended input
- 0 or 12 dB pre-gain
- Selectable AC- or DC-coupling
- Ground lift
- Selectable output impedance (0.1, 10 or 100 Ohms)
- Selectable output stage bias current (30, 130, 220 or 350 mA)
- Disconnectable DC Servo loop

THD is (very pessimistically) less than -120 dB@25 Ohm, 3V rms
No any traces in audio spectrum associated with multiple smps's or control board with high current display. I don't show the spectrums because my measurement system (E-MU 1212m) doesn't feel the difference with or without Z1

A volume control is based on logarithmic step current mode attenuator (equivalent resistance is 3KOhm) with inverting buffer. Relays are special ruthenium Coto 2342. (see LSCM_attenuator)

All signal commutations are with Coto 2342 relays.

A power amplifier is a composite amplifier. (see HA_PA)
Output step response (Rload=50Ohms, 16Vp-p):

A power supply is a multichannel switching mode power supply based on Traco converters.

The amp is controlling by microprocessor board with Osram matrix LED display readable from 3 meters. A user control is two Alps encoders.

Despite the highest objective characteristics, I would like to say a few words on the subjective characteristics.
The sound of amp is light, dynamic and very detailed. Particular attention is drawn to the ability to accurately reproduce large compositions, without any hint of dirt or innuendo. In church choirs individual voices are understandable. Orchestral tutti is so vivid and deep that you are forced involuntary shudder. The sound of the snare drum - it seems that it has sounded in the room next to you. Fans of harp will be encouraged by the energy and rich tones.

I'm listening to it with Denon D7000 and my own DAC (pcm1704 based).

Enjoy.

Meet Oriel Z1, it is made in Russia.

Amplifier PCB

Back side

An externally hosted image should be here but it was not working when we last tested it.

Open cover

An externally hosted image should be here but it was not working when we last tested it.

 

[Telstar]

A power supply is a multichannel switching mode power supply based on Traco converters.

Why?

 

[Nisbeth]

Looks very nice. One question: What is the socket that you are using for DC in?

/U.

 

[rjm]

Thanks for sharing the design.

I like it, it's very Russian:

i.e. "whimsically overboard"

 

[Electrovoicer]

Why?

because classic linear power supply with similar characteristics (regulation, output impedance, impedance vs. frequency) has 10x size, 30x weight and 2x price
There is only one problem - how to cook smps in right way due to switching noise. But proper filtering and placement bring the win.
Three isolated regulated DC/DCs (two for L/R channels, one for digital part) allow me to find optimum grounding scheme.
Another useful thing - the quality of primary power supply is not so important as with the classic linear power supply.

 

[Electrovoicer]

Looks very nice. One question: What is the socket that you are using for DC in? /U.

Thank you.

It is WEIPU connectors the waterproof connectors stand out
Chinese, but very good quality comparable with Amphenol/Tyco/etc.

 

[Electrovoicer]

Thanks for sharing the design.

I like it, it's very Russian:

i.e. "whimsically overboard"

I hope I understand you, thank you.

Be careful with values in composite amp. On schematics you can see only approximated values, which are not guarantee the stability of circuit. In other words, I show only design idea. This design very depends on "parasitic" pcb elements and topology.

 

[qusp]

very nice and very nice choice of parts, panasonic pps around multifeedback loops in the center, I use these caps and love the result. i'm quite interested in the linear post regulation PSU schematic if possible?

nevermind I found the reed relays at digikey

but i was actually mistaking the relay part you mention as being a bank of Wima film caps haha. the ones i'm interested in are the smaller coto part 1042; actually i found them just now too, but only the PTH version, do you have a part number and supplier for the SMD version you apear to have used?

 

[Electrovoicer]

very nice and very nice choice of parts, panasonic pps around multifeedback loops in the center, I use these caps and love the result. i'm quite interested in the linear post regulation PSU schematic if possible?

nevermind I found the reed relays at digikey

but i was actually mistaking the relay part you mention as being a bank of Wima film caps haha. the ones i'm interested in are the smaller coto part 1042; actually i found them just now too, but only the PTH version, do you have a part number and supplier for the SMD version you apear to have used?

hello, thank you!

panasonic PPS are in DC servo integrator thanks to low dielectric absorption and excellent AC performance.

I use two types of Coto relays - 2342 for attenuator/input commutation and 9903 for output circuits/ground lifting which you have seen as Wima caps

All the components in this unit you can find in Digikey and Mouser...

Surprise : there is no linear post regulators in analog part, only proper filtering after dc/dc but it doesn't interfere to reach the necessary characteristics

 

[rjm]

Surprise : there is no linear post regulators in analog part, only proper filtering after dc/dc but it doesn't interfere to reach the necessary characteristics

Well, this is strictly the correct way I would say. No point in using a regulator.

By the way, how difficult was it to get the circuit to run stable, and how did you check?

 

[qusp]

ahh ok, well the schematic is more incomplete than I thought =), i'm not going to try and decipher it further. I thought you had mostly left out the power supply and by post regulator I mean I figured there might be more different and perhaps lower voltages needed than marked on the header for various relay drivers, comparators, logic etc, not so easy to follow multi layer PCB without schematic.

I had mistaken the soic8 and surrounding parts around C20, C12 etc (the only parts I can easily see) and similar collections, in particular around DA8 and DA15 as opamp and comparator based linear regs after composite CL filters.

but I see its more complicated than that =D

 

[Electrovoicer]

By the way, how difficult was it to get the circuit to run stable, and how did you check?

I have started from SPICE model with general "parasitic" elements like capacitance across resistors, inductance across capacitors etc. and have found good estimation which guarantee good phase margin. Also I studied the sensitivity of schematic for the change of values, including "parasitics".
On practice, 30% increasing of only one capacitor has need to reach necessary conditions (comparing with model). You can see transient response curve of real amplifier on picture above.
In common case, "composite" topologies are difficult to tune due to complex impact of one parameter on another. Especially if we want high amplification at the internal feedback loop, more than the general amplification of "composite" amplifier.
This approach allows us to reach very high parameters - the "error" amp is working in small signal mode. So you can choose high precision low noise op amp. In this case overall characteristics combine the precision of input amp and speed/current of output amp.
The use of inverted topology give us additional benefits. In this case "error" amplifier is working with zero common mode voltage. This trick prevents "capacitance modulation" distortions in input stage of "error" amp. There is only one limitation - noise. So you can choose optimum input impedance. Very low input impedance would load the output of previous stage. High input impedance would be the source of noise (it would multiplied on input current noise of op amp). My opinion for most cases - impedance must be few kilo-Ohms for bipolar op amps.