Fosi Audio ZA3 Amplifier Review

[m0nsieur]

Frequency response of the Fosi ZA3 (48V) vs my old Marantz SR4021 while driving a Polk R200 bookshelf speaker.​

Disclaimer: this is my first attempt at posting measurements.​

Test setup​

Room: 13 x 16 ft untreated + a stairwell, kids toys all around the room.
Speaker location: center along the side of one of the 16 ft walls, sitting on top of an entertainment console (full of kids toys), 8in away from the rear wall (not ideal I know).
MiniDSP Umik-1 location: along the opposite 16 ft wall, perched above the top of a couch and aimed at the speaker.
14 AWG oxygen free copper speaker wires with banana plugs.

Methodology​

Output level matching: I used REW to output a 1000Hz test tone, and (with speaker connected to the amplifier) connected a multimeter + alligator clips to the + and - speaker connectors going out of each amp. I turned up the volume until the multimeter read 5V AC. This was about the 4 o'clock position on the Fosi ZA3 (not far from max volume) and the -20dB level on the Marantz SR4021.

I heard a short buzzing resonance during the frequency sweep when the output was set to 5V. I did not try to locate and suppress the buzzing. I'm assuming that with both amps the same buzzing was measured. If not, that might explain the peak around 6500 Hz for the Fosi.

Summary​

It looks like I should be more concerned with modes and reflections than I should be with the frequency dependence of the Fosi ZA3.
According to this thread, these speakers have an impedance peak of 20 ohms around 1.5 KHz, but I don't see any difference in SPL at that region with the Fosi.
I don't know what the load impedance, if any, there is of the Marantz SR4021, but for this comparison I'm assuming it is class AB and has a flat impedance response.

Other thoughts​

• I ordered the Fosi on their website. It took about a month for them to ship, but once shipped it arrived pretty quickly,
• I like the compact size of the Fosi.
• The speaker terminals are as slippery and cheap as all other budget amps. Use banana plugs!
• I don't like the switches on the front of the Fosi because they can be toggled (by little kids or whoever).
• The Fosi builds volume quickly when connected to my CD player's line level outputs. Probably due to the high input gain on the ZA3 that @amirm mentioned. The range of the volume knob is more gradual with lower level sources (computer built-in sound output, Apple USB-C dongle, older LG phone with quad DAC).
• I want to build a set of Linkwitz LXMini speakers and am still considering using a pair of ZA3s as budget amplifiers (even after reading the Aiyima A70 review).
• I slightly favor the sounds of the Marantz, but I could be biased or just used to the sound signature of the Marantz.

 

[amix]

I can't understand the filtered sub out.

It's about keeping the low frequencies away from the mid-bass driver on 2-way speakers.

 

[solderdude]

What if the op amps are the reason for the 2dB like Greg says? It's a valid test then because changing the op amps and leaving some out caused that.

There is no what if.

Op-amps do NOT determine the gain of an op-amp circuit. ONLY the components around it do and these do not change when changing op-amps.

 

[iLoveCats]

There is no what if.

Op-amps do NOT determine the gain of an op-amp circuit. ONLY the components around it do and these do not change when changing op-amps.

I think it deserves testing before dismissal. There's definitely something going on in that video, I hear it with the built in speakers under the keyboard on a MacBook Air.

 

[Doodski]

Op-amps do NOT determine the gain of an op-amp circuit. ONLY the components around it do and these do not change when changing op-amps.

I am pretty rusty on OP amps this late in the game but yes the peripheral gain setting components are a constant and if a higher gain OP amp is inserted the gain remains the same?... dang I wish I had my old textbooks... Never have I seen a better OP amp textbook that the one I had and learned from by Floyd. I thought the gain would change very slightly... But you are the resident expert and an actual Working Electronic Engineer here so I accept you are de factO correct... I've never known you to be anything other than correct so I will file this info under new and correct info regarding OP amps.

 

[somebodyelse]

I think it deserves testing before dismissal. There's definitely something going on in that video, I hear it with the built in speakers under the keyboard on a MacBook Air.

No, it really doesn't. It's part of what makes an operational amplifier - the 'operation' goes back to analog computing. The operation (multiplication, addition etc.) is defined by the feedback network. If the amplifier changes the gain then it is _not_ an operational amplifier.

 

[solderdude]

I think it deserves testing before dismissal. There's definitely something going on in that video, I hear it with the built in speakers under the keyboard on a MacBook Air.

I have been designing analog circuits for decades using op-amps. You should consider that op-amps have gains in the ten thousands to millions (100dB open loop gain is not uncommon) and it really does not matter which one you use in a circuit that has the feedback components set to say 1x to 10x gain. The result is ALWAYS the same and determined by the gain resistors. What does change is the bandwidth of the circuit and distortion in the highest frequencies when the gain is higher.

The difference in gain between the sections in the recording is NOT caused by the op-amps but of other causes (the recording and the flawed method used).

 

[312elements]

No, it really doesn't. It's part of what makes an operational amplifier - the 'operation' goes back to analog computing. The operation (multiplication, addition etc.) is defined by the feedback network. If the amplifier changes the gain then it is _not_ an operational amplifier.

So what’s your conclusion? Are you choosing to ignore the difference. Are you suggesting there isn’t a difference? Are you suggesting it must be broken ? Are you suggesting the recording is doctored? Enlighten us please.

 

[DanTheMan]

I have been designing analog circuits for decades using op-amps. You should consider that op-amps have gains in the thousands to millions (100dB open loop gain is not uncommon) and it really does not matter which one you use in a circuit that has the feedback components set to say 1x to 10x gain. The result is ALWAYS the same and determined by the gain resistors. What does change is the bandwidth of the circuit and distortion in the highest frequencies when the gain is higher.

The difference in gain between the sections in the recording is NOT caused by the op-amps but of other causes (the recording and the flawed method used).

So IOW, the guy recording the video shouldn’t have to ‘level match’ at all when he switches the op amps?

 

[solderdude]

So IOW, the guy recording the video shouldn’t have to ‘level match’ at all when he switches the op amps?

exactly.
Of course it would be important that the volume control and acoustic conditions did not change either.
In order to change the op-amps the amp has to be taken apart and who knows what else changed.
When he had 2 amps and one has different op-amps there can be different volume control settings for instance.
Doing the incorrect 'level matching' is the biggest error. On the other hand if that were done perfectly and there would be no tells (knowing when it was switching o what) I am sure no one would be able to reliably tell the difference.

Best would have been to record the electrical signal (unchanged volume control) coming from the speaker outputs and put up the files separately and use ABX tester to determine if you can tell the difference reliably.

In any case the level difference (which is audible) is a give away that there is something fishy with the comparison as is the other tell namely you can see when the switch is and what you are listening to. slightly louder is always perceived as 'better' but not necessarily louder

 

[somebodyelse]

So what’s your conclusion? Are you choosing to ignore the difference. Are you suggesting there isn’t a difference? Are you suggesting it must be broken ? Are you suggesting the recording is doctored? Enlighten us please.

I'm saying that if the Sparkos changes the gain then it is by definition not an opamp. This seems the less likely explanation of a difference in level in the recording. The more likely is that the level changed by other means (signal level, volume control moved, mic moved etc.) The charitable interpretation is that it was never checked.

 

[TedBaker]

Perhaps I'm being stupid so please forgive me if I am. But I bought a Wiim Amp which I use for the TV room downstairs and a Fosi ZA3 for my office to replace Cyrus amps. Love the Wiim Amp. However the Fosi Amp which I am using via a SMSL DAC and a Wiim mini sounds very hissy in the top range. I started by turning the amp up to max and adjusting volume on my PC, but there was an audible hiss and a bit of distortion from the tweeter on the quiet tracks. Nothing from any of the other speakers in the towers I am using. It is usable for me, because I am in a relatively nearfield situation and I can turn the amp to two thirds and I will have enough volume, but it is a bit disappointing. Tried to replicate this using the Wiim Amp and it was clean. I thought I'd damaged the tweeters somehow but that wasn't the case.

 

[Petevid]

Perhaps I'm being stupid so please forgive me if I am. But I bought a Wiim Amp which I use for the TV room downstairs and a Fosi ZA3 for my office to replace Cyrus amps. Love the Wiim Amp. However the Fosi Amp which I am using via a SMSL DAC and a Wiim mini sounds very hissy in the top range. I started by turning the amp up to max and adjusting volume on my PC, but there was an audible hiss and a bit of distortion from the tweeter on the quiet tracks. Nothing from any of the other speakers in the towers I am using. It is usable for me, because I am in a relatively nearfield situation and I can turn the amp to two thirds and I will have enough volume, but it is a bit disappointing. Tried to replicate this using the Wiim Amp and it was clean. I thought I'd damaged the tweeters somehow but that wasn't the case.

Have you fixed the volume output of the Wiim Mini at maximum level in the app or are you using variable volume in the Wiim to adjust volume?

 

[TedBaker]

Have you fixed the volume output of the Wiim Mini at maximum level in the app or are you using variable volume in the Wiim to adjust volume?

I have the volume at about 60% on the Wiim / PC. I get the hiss when I turn the Fosi amp to maximum. But I think the hiss comes at that end of the chain with the Fosi amp. I'm using the optical output of the Wiim.

 

[tinnitus]

I use the ZA3 after a Topping DM7 with maximum Vol and have no hiss or hum.

 

[TedBaker]

I use the ZA3 after a Topping DM7 with maximum Vol and have no hiss or hum.

I am wondering whether it is the combination of the amp and the speakers - B&W 683 s1 - they're old, but they've not really let me down. It's only the tweeter, none of the other three speakers in the tower are hissing when the Fosi amp is on full volume. I replaced an old Cyrus amp rated at 70w as a test to see how good the Fosi amp was. I had no real issues with the Cyrus which I used on preamp bypass from the SMSL DAC and has been going from 2006 when I originally bought the speakers. I thought the hiss might be interesting in the context of the Fosi driving an older speaker and the way it handled the load, but I might be getting my wires crossed. It sounds perfectly fine when I turn it down to 2/3s with no hiss. I thought the Fosi might drive the speakers better than the Cyrus amp. The amp was only £150 so it's not a big deal. At least I am saving a bit on the electricity bill.

 

[iLoveCats]

I have a little hiss if you put your ear on it with some 99db sensitivity compression horns. No hiss with a PA5 amp previously.

 

[Jim777]

Does the load dependency have a linear effect on output levels (same for any power level), so that EQ can get rid of it for a specific set of loudspeakers?

 

[somebodyelse]

Does the load dependency have a linear effect on output levels (same for any power level), so that EQ can get rid of it for a specific set of loudspeakers?

Yes (ignoring second order effects like the effect of temperature on impedance)

 

[YesChickenNuggets]

MiniDSP Umik-1 location: along the opposite 16 ft wall, perched above the top of a couch and aimed at the speaker.
14 AWG oxygen free copper speaker wires with banana plugs.

Summary​

It looks like I should be more concerned with modes and reflections than I should be with the frequency dependence of the Fosi ZA3.
According to this thread, these speakers have an impedance peak of 20 ohms around 1.5 KHz, but I don't see any difference in SPL at that region with the Fosi.
I don't know what the load impedance, if any, there is of the Marantz SR4021, but for this comparison I'm assuming it is class AB and has a flat impedance response.

I am not familiar with microphone specifications, but does the umik have some sort of accuracy delta +/- xxdb ? If these specs are usually outside of +/- 0.5 db, Can we even use the microphone for this test ?

Also, are these microphones specced for a specific temperature and humidity ? I'd assume a large deviation could easily cross 0.5 db no ? Is frequency sweep the proper way to measure what small difference there may be ?

Could the temperature of the voice coils also not easily change sensitivity to cross 0.5db .

I'm not criticizing you m0nsieur, I'm just wondering outloud. Overall, it's pretty clear linearity of the physical driver is several orders of magnitude worse than the amp, maybe putting us back to the golden ears side of audiophilia discourse.