trinityamps.com

Since Stephen pointed out the Tone Stack Calculator (see this thread), I've wanted to take some measurements to see how theory and practice combine. Actually, I've been doing all kinds of diyAudio projects for many years, and wanted to do frequency response measurements, but never ponied up for the hardware. When I built my Triwatt, I heard an obvious, immediate difference in tone compared to my other amp, and wanted to see if I could correlate what I hear to the measurements.

What follows are some graphs of frequency response measurements I did with my Triwatt. First thing to note: this is literally the first time I've taken these kinds of measurements. I just received the hardware yesterday, and this morning, started clicking buttons in the program until it appeared to do what I want. So what follows is very much a first rough-pass on the frequency response measurements - take them with an immense grain of salt. When I have more practice and experience with the tools, I will come back and repeat these measurements. This is really just a "teaser" to whet your appetite!

A quick rundown of the tools:

• Microphone: Behringer ECM8000
• Audio Interface: Focusrite Scarlett 2i2 3rd Gen
• Software: Room Equalization Wizard (REQ) v5.19

First, as a crude initial sanity test, I did a measurement of my stereo speakers, the Solstice MLTL. These are DIY speakers, designed by Jeff Bagby, with a fairly flat frequency response. Played through a Topping D70 DAC and Modulus-86 amplifier (perfectly flat modeled and measured performance). I put the mic a few inches from the speaker, roughly centered between the tweeter and top woofer. For what is basically my first measurement ever, I think it looks reasonable. You can see a slight dip between 1kHz and 2kHz, and this speaker's crossover point is 1.7kHz, so that lines up nicely. Other aberrations I'm chalking up to room effects and inexperience with the tools.


Some notes about how the Triwatt was setup for these measurements:

• The master volume was around 8:00 or 8:30
• Overdrive off
• All other knobs at noon, unless noted otherwise
• I recently modified my Triwatt C6 value to 10nF, as opposed to the default 1nF. So keep that in mind, mine is not a stock Triwatt, and per the TSC modeled response, this makes a significant difference (Triwatt C6 is C2 on the TSC site).

The first two graphs are just for comparing normal versus bright versus link inputs. The first graph is just normal versus bright, and the second adds the link input.




All subsequent measurements were taken from the link input. The next three images are min versus max settings for each of the treble, middle and bass knobs.







Again, keep in mind, this is all very crude, as I'm just learning how to use these tools. The room effects were not taken into consideration, nor was any of the equipment calibrated. But I think these graphs at least give a sense of things.

Lastly, here is my crude mic placement. I don't even have a mic stand, so I used the mic's case as a makeshift stand!

Back with some more measurements.

I now have a proper microphone stand, so I can position the microphone consistently.

I've taken a handful of measurements since my initial posting, and found that the process is very sensitive. I was initially doing these measurements in my office, which is not acoustically treated and fairly small (about 10 x 11 feet). The only "treatment" is an area rug, but all other surfaces are hard (wood or glass). Most would call this an acoustically poor room.

So in the measurements that follow, I moved the speaker cabinet out into the foyer of my house. This is one of the larger, more open areas of my house, so I'm hoping it reduces the room effects to some degree.

I also realized this REW software has a button to average multiple response measurements. So, I took all measurements three times, and averaged them. The graph that follows has three measurement curves. Each curve represents a different speaker paired with the Triwatt at the same settings. The microphone is about 1 inch from the grille cloth. I tried to center the microphone at the speaker's center (though I eye-balled this, can't claim any precision). Triwatt configuration is as follows:

• JJ 6V6 power tubes
• New production Tung-Sol 12AX7 in v1
• Ruby 12AX7AC7 HG in v2, v3 (supposedly the same as TAD 7025-S, Preferred Series 7025 and Gold Lion Standard pin 12AX7)
• Sovtek 12AX7LPS in v4 (phase inverter)
• All components in the circuit are per the build documentation (recall, I had a cap in the tone stack swapped out in my previous measurements, that cap has been reverted back to stock value)
• Overdrive off (and turned down to 0)
• Master volume at 9:00
• All other controls at 12:00
• Using Link input

The three speakers tested are as follows:

• Reeves Vintage Purple (green line) - this is in the 1x12 Triwatt combo cab that Trinity built for me
• 12" Eminence (blue line) - not sure which model, this is a Blackheart 1x12 cab that I got many years ago to mate with a Blackheart BH5H cab
• WGS Retro 30 (orange line) - this is in a 1x12 combo cab built by Sour Mash; it currently houses that speaker plus a Ceri@tone Dizzy30 amp

I'm curious how big of a difference can be seen in the frequency response of different amps with the same speaker. To my ears, the Dizzy30 and Triwatt sound very different. Spoiler: their frequency response graphs are very different. This is all for the ef86 channel of the Dizzy30. I used the half-power setting, "high" input, preamp gain at noon, cut at noon, and master around 8:00. The only other control is the tone knob, which is a 6-way switch that changes the coupling capacitor between amp stages (between ef86 and phase inverter IIRC). I took two measurements, one at each extreme of that switch:



What follows is the same graph, but add the response of the Triwatt with the WGS Retro30 speaker (orange line). In other words, the cabinets and speaker are the same, but the amps are different.



Last is just a repeat of the Triwatt with the Reeves Vintage Purple speaker in isolation (for better visibility):



I'm curious what others think of this. My biggest surprise is how ragged it is from 1 kHz and up. My expectation, based purely on Internet hearsay of people calling the Hiwatt "hi fi" and sometimes even "sterile", I would expect a much flatter and even response in the range where the human ear is most sensitive (say 1kHz to 3 or 4 kHz). But this is definitely not a graph of hi fi equipment! Not saying that's a bad thing, not by any means - just saying it's different than what I would have predicted. I suspect it's the ragged frequency response that gives any guitar amp its signature tone.

Eager to hear others weigh in on this!

Matt, checked the Solstice link. Those are nice speakers!

As far as the measurements go, are you planning to do any off axis measurements? Also, does your test software have a function to allow you to adjust for room acoustics? That might give you additional data. I do like the fact that you are making your measurements in a consistent manner.

What is interesting is 1) the big notches at 700 and 800 mHz from the Eminence (speaker bound?), 2) the fact that all three have approximately the same notch at 1K (amp bound?) and 3) the Vintage Purple has the flattest trace while the Eminence and WGS Retro have the pretty much the same trace at 10K and above (speaker bound again?). In any case, it's all interesting.

_________________
"The only problem with being retired is you never get a day off"

I really like them. The kit from Parts Express is a good value, especially when you consider the cost of the individual drivers. They were designed by Jeff Bagby, who is a legend in the DIY speaker community. Sadly, he recently passed due to coronavirus.




I wasn't planning on it, but that doesn't mean I won't at some point. In my mind, off-axis measurements would be more in the interest of the speaker and/or cabinet. With microphone-based measuring I'm doing, obviously the speaker+cab+room plays a huge role, so I'm implicitly measuring them. But, at least as far as the Triwatt is concerned, my primary goal is to measure the amp itself, and compare its behavior with different speakers. As I typed that, I realized I need to figure out way to do a direct measurement of the amp's output, so as to eliminate the speaker+cab+room from the equation.




It does! In fact, that is the software's primary use-case. It's "REW", Room Equalization Wizard, and it's specifically designed for tuning your hi-fi or home theater speakers in the room. It's very powerful software, and I've literally just scratched the surface of what it can do. I basically loaded it up and started clicking until I got measurements!




Thank you! Yes, I had similar observations as well (didn't want to say too much initially, was hoping to spark some more discussion). That said, I suppose a co-equal primary goal here is to actually become competent in taking measurements like these. Meaning, we can't read into the measurements too much at this point, given that I'm very much an amateur at this.