Monday, February 19, 2018

FPB vs Trans Nova

The FPB 300 right channel now seems to overheat at low playback levels regardless of input or speaker connected.  One time in January it reached 207F before I shut it down.  That time it had reached 197F during playback, I shut the music off, and the temp kept rising anyway for ten minutes.  Power consumption was about 900W.  This looks like at least one malfunction!  But it could have been noise or DC offset at the input.

After the 207F incident, I reversed input and output channels, put in place a brand new DAC, and connected the RCA instead of XLR inputs, with shorting plugs in place.  So I changed about everything that could be changed.  I resumed playing a test with maximum 2W playback levels as measured by my Keithley meter on sine wave tests (see below).  It seemed fine until after about 30 minutes consumption hit 1000W.  It just kept on rising to 1380W when I shut it down.  So, putting the right amp channel on the left speaker didn't help.  But was it the right speaker that was now causing a problem with the left channel?

For that test I shorted the XLR input of the right channel with a Cardas XLR shorting plug, disconnected the output too, and played the left channel of the amp with the right speaker as before.

That is operating flawlessly for hours now, total consumption around 530W (right on for just one channel playing, the other idling).  Showing that "the problem" didn't move along with the right speaker.  The left channel plays the right speaker just fine, but apparently the right channel goes haywire with either the right or left speakers connected to it playing low level signals, stays cool only with shorted input.

I will be filing a problem report soon.

Meanwhile, while doing this high stakes testing I tested another hypothesis: that the Krell sounds more dynamic and layered because it has higher damping factor, especially at high frequencies.

My actual measurements below show the reverse is true!  The Hafler 9300 has less voltage drop under load at middle frequencies AND even more so at higher frequencies!  In the low power test I performed.

Here is the raw data:

Hafler 9300 connected to right Acoustat
1kHz 1.765V
16kHz 0.213V

(The actual test files apparently are rolled off at higher frequencies deliberately.  Emotiva level is -4dB.)

Hafler 9300 not connected
1kHz 1.802V
16k 0.225V

Krell FPB 300 connected to right Acoustat (left amplifier channel, Emotiva at -1.5dB)
1kHz 1.723V
16k   0.204V

Krell FPB 300 not connected
1kHz 1.779
16k    0.220

There are many ways to compare these few numbers.  Let's compare like with like, at 16kHz (the most stringent test).

The Krell shows 0.656dB loss under load relative to open at 16kHz.

The Acoustat show 0.476dB loss under load relative to open at 16kHz, an almost 0.2dB difference.

At 1kHz, the Krell vs Hafler numbers are 0.28dB and 0.18dB, a 0.1dB difference.

Now, curiously, under no load, the Krell is about 0.2dB flatter, so the effect of the extra 0.2dB loading loss is to make their response curves equal under load!

Because of that, I'd expect the effect of the apparently slightly poorer high frequency damping factor (or output Z if you prefer) of the Krell (as wired, but actually the Hafler has much longer speaker cables of the same gauge...so that doesn't explain this difference either) not to have an audible effect on high frequency response.

So the two parts of the hypothesis are both rejected by these data.

The Krell shows no audible superiority, and since I'd consider better HF damping to be more important the flatter unloaded response, the Hafler looks to be the objective winner here in one tiny way.

Since only one channel is safe to run, I did blind AB test on just the right speaker, Krell vs Hafler.

In about 10 minutes of A/B switching, I hear no consistent difference, with Krell at -1dB and Hafler at -4dB (less than 0.5dB advantage to Krell, with -1.5dB being the other way, as shown by numbers above).