Topic: Garage 1217 Project Sunrise III measurements (Read 1987 times)

atomicbob · « **on:** May 25, 2015, 01:07:02 AM »

This hybrid tube - mosfet ampifier has been providing me with considerable enjoyment for some time now. Here are some measurements I made today in a quest to dig deeper into why this amp sounds so good to me. The distortion is particularly good considering the low voltage plate tube front end and the no feedback topology. The listening station is a Mousai MSD192 (measurements found here), the PS3 and HD6x0 or D5000 headphones as the cans with most audition time on this system. I should think there would be more discussion of this amp given it has an exceedingly high performance/price ratio.

The data presented were collected as follows:

1. PrismSound dScope III, picoscope 5243B
2. DAC unbalanced output RCA
3. 100 Kohm load used for measurements
4. 44 KHz sample rate, 24 bit depth
5. USB input – Audioquest Forest
6. Unbalance cable DH Labs Silver Sonic Air Matrix RCA
7. Vaunix Lab Brick USB hub
8. PS Vane 12AU7 used in PS3
9. PS3 output set for 32 ohms

Picture of the Kitchen Counter Audio Lab

Tests were run at a gain consistent with my personal listening sound levels which is between 70 to 77 dBC SPL.

A 0 dBFS sinusoid at that gain is 700 mVpp:

There is a 1 dB imbalance between the channels at my preferred listening level. Not enough to concern me for recreational listening.

Square waves can reveal a lot about an amplifier's performance. For reference morphology, the signal generator waveform looped back to the oscilloscope input at 1 KHz, 700 mVpp.

Signal generator loop back to oscilloscope input for reference

PS3 square wave 20 Hz 700 mVpp 10 mS / div 100K load

PS3 square wave 20 Hz 700 mVpp 50 uS / div 100K load

PS3 square wave 20 Hz 700 mVpp 10 mS / div 50R load

PS3 square wave 20 Hz 700 mVpp 50 uS / div 50R load

Noting the Mousai MSD192 has an excellent 20 Hz square wave resp onse and so too the PS3, we have a theme developing.

atomicbob · « **Reply #1 on:** May 25, 2015, 01:08:04 AM »

Part B of measurements.

Here are 20 Hz square waves with an HD650 as the load:

PS3 square wave 20 Hz 700 mVpp 10 mS / div HD650 load

PS3 square wave 20 Hz 700 mVpp 50 uS / div HD650 load

dScope measurements

PS3 amplitude - phase - gain

PS3 signal to noise

PS3 THD+N THD nth harmonic distortion

Three things to note about the distortion measurements.
1) Distortion will be dominated by the tube chosen for the input gain stage.
2) The 3rd harmonic distortion is 20 dB lower than the 2nd harmonic distortion. My ear finds even harmonic distortion pleasing and odd harmonic distortion offensive. Thus the PS Vane 12AU7 is behaving the way I prefer in the PS3.
3) The PS3 does not use feedback to lower the distortion specs

editorial - I believe there is a tendency to focus too much on THD numbers rather than look at distortion spectrum and type such as even vs odd in an attempt to reduce a complex topic to a single number representing a sort of goodness rating.

atomicbob · « **Reply #2 on:** May 25, 2015, 01:09:15 AM »

Part C of measurements

If you made it through all of the numbers and graphs, you can see why I think this hybid headphone amp rocks.

Ringingears · « **Reply #3 on:** May 25, 2015, 02:23:28 AM »

Yes you would think there would be more discussion of Garage1217 amps in general.

Solderdude · « **Reply #4 on:** May 25, 2015, 07:55:40 AM »

For those that like squarewave shots; a 10kHz and 100kHz squarewave (with 60 Ohm load)
(The 'noise' you see on the signal is from the scope, not there in reality)

Do note that L-R balance, Frequency Response, THD and harmonic character are all tube dependent as Bob already mentioned.
L-R balance differences I have seen vary between 0.1 dB and 2.1 dB with different tubes (all pulls).

Using an (old) Philips ECC88 tube and 32 Ohm load I measured FR: -0.5dB: 10Hz - 150kHz / -3dB: 4Hz - 350kHz
With tubes like the ECC83 the FR can easily be reduced to -3dB: 5Hz - 15kHz so the FR range can vary substantially.

The amount of distortion greatly depends on the output level.
The lower the output level the lower the distortion.
At 'Normal' listening levels the THD levels are considerable lower b.t.w.

The schematics of SR-III are found here:
http://www.mediafire.com/view/06g72zyu2d32gp6/Sunrise_3.1_schematic.pdf

atomicbob · « **Reply #5 on:** May 25, 2015, 02:45:22 PM »

Quote from: Solderdude on May 25, 2015, 07:55:40 AM

The amount of distortion greatly depends on the output level.
The lower the output level the lower the distortion.
At 'Normal' listening levels the THD levels are considerable lower b.t.w.

An excellent point! The dScope measurements were performed with a -1 / -2 dBu output. I tend to listen with a level averaging -10 dBu or less. I should reprogram the dScope amp scripts to allow a lower reference and repeat.

Tube dependency allows tuning amplifier character to suit one's tastes. It also makes comparison of notes between listeners a little more difficult as they need to discuss which tube was used that gave them the sound they heard.

I did not put up the crosstalk measurements as I found considerable variance depending on the cable used for TRS to stereo RCA for connection to the dScope input. The least expensive Radio Shack performed much better than a major name high end super duper TRS to stereo RCA cable, for example.

atomicbob · « **Reply #6 on:** May 26, 2015, 08:34:34 PM »

Conditions:
1. PS Vane 12AU7
2. PS3 32 ohm output setting
3. 50 ohm load
4. 4 Vpp input from signal generator (+5.23 dBu)
5. 600 mVpp output from PS3 into 50R load

Frequency Response

-0.5 dB 3 Hz to 70 KHz
-3.0 dB 1 Hz to 200 KHz

10KHz and 100 KHz square wave response
top trace - signal generator input
bottom trace - Project Sunrise III output

10 KHz square wave response:

100 KHz square wave response:

100 KHz square wave response x-axis zoom:

Cos · « **Reply #7 on:** September 14, 2015, 09:17:19 AM »

Interesting page! Thanks for taking the time to do the measurements!

Wondering how the Sunrise compares against the local classic, the much revered Bottlehead Crack. Using just one (exotic) tube, to me has the advantage of beeing cheaper than two for Crack, not to mention four for Valhalla 2 (finding matched tubes EDIT: other than the smart and nicely priced set provided by Shiit - extra headache).

Also wondering if the Sunrise/Horizont schematics can be modified similar to Valhalla 2 to be optimized for both high and low impedance headphones.

Solderdude · « **Reply #8 on:** September 14, 2015, 12:56:34 PM »

Quote from: Cos on September 14, 2015, 09:17:19 AM

Also wondering if the Sunrise/Horizont schematics can be modified similar to Valhalla 2 to be optimized for both high and low impedance headphones.

The amplifiers differ too much to compare.

In case of the Valhalla2 there are 2 output resistances to 'choose' from. 14 Ohms or 3.5 Ohms.
The reason there is some difference in output R has to do with the topology (the way the circuit is built) and the amount of feedback.
It's an 'all tube' design.

More feedback = lower gain = lower output R
Less feedback = higher gain = higher output R
Those aspects are directly coupled within the design.
It also means that even though the valhalla can be used with low impedance headphones it is more suited for higher impedance headphones.

For the SR/Horizon things differ.
Firstly this is a (low voltage) hybrid design (more akin to Lyr than Valhalla, EDIT: in output stage)
The gain and output Resistance can be set separately and are not coupled.
The internal gain of the amplifier is not determined by the amount of feedback (which it is in Valhalla)
In SR/Horizon the gain is determined by the tube used and there is NO overall feedback at all.
The input signal can be attenuated (so not gain control technically but attenuation control).
The amount of attenuation can be changed quite easily by the user (SR-III, Horizon III).
For the older versions the 2 gain steps are fixed.
The output R can be set in 3 steps: 1.5 Ohm, 35 Ohm and 68 Ohm for SR and 1.5, 35 Ohm and 120 Ohm for Horizon.

Both SR and Horizon can be used for low and high impedance headphones.
the SR can provide a bit more current in low-Z headphones while the Horizon can provide a higher voltage for higher Z headphones or low Z headphones that could best be driven from 120 Ohm.

schiit · « **Reply #9 on:** September 14, 2015, 02:10:10 PM »

Quote from: Solderdude on September 14, 2015, 12:56:34 PM

Firstly this is a (low voltage) hybrid design (more akin to Lyrr than Valhalla)

Incorrect. Both Lyr and Lyr 2 are high-voltage hybrids, not starved-plate designs. They have a proper 200V B+ rail. Which may make them unique at their price points.

Author Topic: Garage 1217 Project Sunrise III measurements (Read 1987 times)

atomicbob

Garage 1217 Project Sunrise III measurements

atomicbob

Re: Garage 1217 Project Sunrise III measurements

atomicbob

Re: Garage 1217 Project Sunrise III measurements

Ringingears

Re: Garage 1217 Project Sunrise III measurements

Solderdude

Re: Garage 1217 Project Sunrise III measurements

atomicbob

Re: Garage 1217 Project Sunrise III measurements

atomicbob

Re: Garage 1217 Project Sunrise III measurements

Cos

Re: Garage 1217 Project Sunrise III measurements

Solderdude

Re: Garage 1217 Project Sunrise III measurements

schiit

Re: Garage 1217 Project Sunrise III measurements

(adsbygoogle = window.adsbygoogle || []).push({});

Author Topic: Garage 1217 Project Sunrise III measurements (Read 1987 times)