Tube Amp Output into Resistive vs Speaker Loads.

[Gnobuddy]

Then, I ran the same baseline plot of 4 ohm Resistive @ 50% Presence and added the response plots with the Test Speaker attached. What a world of difference that shows!

Thank you for the measurements and pics!

Having grown up with semiconductors, it only recently dawned on me just how high the output impedance of a no-feedback, pentode-output guitar amp would be. In response to my theoretical guesstimates, diyAudio member Printer2 did some actual measurements on a couple of his DIY guitar amps, and his numbers translated to 40 ohms output impedance for one amp, and 50 for the other.

Either of those numbers is so much higher than the DC resistance of a speaker voice coil, that it amounts to driving the speaker from a current source.

From theory, a loudspeaker (unmounted, or in an open-back cab) fed from a source impedance considerably higher than the voice coil resistance will have its on-resonance mechanical Q rise to virtually Qms. Driven from a zero-impedance source, its mechanical Q on resonance is Qts. For most speakers, Qms is far, far higher than Qts. Expect a massive peak down there, in other words.

But your data is the first time I've seen actual measurements of this. We can even see the port resonance of the enclosure at around 15Hz, and the speaker fundamental resonance at around 60 Hz.

I am a little surprised that we're not seeing a treble boost due to the rising reactance of the speaker voice coil. Perhaps this particular speaker has a low-inductance voice coil (copper pole cap, etc)?

Thanks again for the measurements!

-Gnobuddy

[Leo_Gnardo]

After first trying to drive the open-circuit 4-Ohm Output Jack of the Mesa Stereo Simul-Class 290 Power Amp from a 1k ohm signal source impedance, and had it break into a large asymmetrical square wave output and pull additional AC mains current, I quickly shut it down (S/B mode) to re-think the procedure.

I've seen Mesa amps do the "Mesa wobble" without a load on the output, as far back as 1987 Mark II. Since then I do my best to make sure there's a load on any Mesa amp's output as it can't be good for the output tubes and may also stress other critical components. Good research nevets, I'm paying attention to what you're doing and thanks for doing it!

[nevetslab]

Actually, the port tuning is at the lowest impedance (around 34-36Hz) between the two LF resonance peaks (looking at the BXE-115HL5 plot). I haven't disconnected the Piezo Tweeter & crossover network to see what that does. It's merely one of the two 15" speaker cabinets I have in the shop for testing purposes (The other is a Hartke 8 ohm box, don't know the model).

Tomorrow, I'll have a look to see how high I can raise the 'load resistor' across the amp output before it takes on the misbehavior I experienced when I tried to feed the output with no load, just that of a 1k ohm sine wave source. Seeing the trend in these latest plots, being high enough in 'loaded' impedance, I'm sure the real source impedance will be considerably higher than the 'loaded' curves show.

When I first saw the very high LF resonance when the Presence Control was at 50% and lower, it was at that same frequency range that I've seen Dual and Triple Rectifier Amps oscillate (motor-boating) unless you engaged the 'Modern' voicing switch, which removed the NFB connection. When I first obtained this Stereo Simul-Class 290 back in 2010 (for $75!), I recall it had that LF oscillation. 4 ohm cabinet with a 6ft cable. Here, Ive been using a 25ft cable, reaching one end of the shop from my test bench.

In the next day or two, I'll be setting up both a solid state amp (probably that Alesis RA-100 used in the setups) and this Mesa amp, match the output drive levels at 160hz, where the Ampeg speaker is at it's 'nominal' 0-phase impedance, then drive each with Pink Noise to see if there is any level difference. I'll set the Presence control at 50% rotation, where it's flattest over the bandwidth. and measure it on my B & K 2035 Signal Analyzer, 1/6 Oct with the calibrated 1/2" mic. Not an ideal room to make acoustical measurements in, but, for crude comparative trend info, both amp/speaker combinations will have the same conditions imposed. It's been a fun and interesting project so far. As time permits, now having worked out a measurement system, I'll make measurements of a number of Fender and Vox combo amps from CenterStaging's vast rental inventory.

I've been away from the Thiele & Small speaker characterization details long enough to have forgotten all those relationships as you were discussing.

Fun in the physics lab.

[Tom Phillips]

What am I missing when looking at the posted graphs? I see various colored lines plotted but I don't find a legend telling me which line represents which measurement. Am I missing something obvious right in front of me? Maybe too tired after a long day of work?

[nevetslab]

The horizontal lines that roll up from the left side and roll off on the right side, straight across from around 50Hz to 20khz...those are my resistive calibration lines. I ran plots of 32 ohm, 16 ohm, 8 ohm, 4 ohm and 2 ohm for the 4-ohm output tap, and 64 ohm, 32 ohm, 16 ohm, 8 ohm & 4 ohm calibration lines for the 8 ohm tap.

Normally, when I ran plots like this on an XY plotter or on my storage scope, I'd make use of the printed graph paper and fine-tune the impedance magnitude range to match the graph paper, and the same with the 8 x 10 scope gratical lines. With this PCSGU250 USB scope-generator, it lacks the refinements of being able to fine-tune plots, so I added calibration lines. As it is, I found the left and right ends of those cal lines, where they roll downward....that's associated with the voltage across the sense resistor. I played with that function on this series of ouptut Z plots, and never did get it to start and end straight across, and ended up leaving it with that roll-off.

All this would no doubt be simplier if I owned an Audio Precision System 2 or System 1, properly loaded.

On the actual impedance plots, as well as the frequency response graphs that I first posted, the annotation added on each graph does call out the variation, and I kept those to just 2 to 4...such as the Output Impedance for 4 ohms tap.....The lowest LF peak curve is with Presence Control at Max, the middle 15Hz resonance peak is 50% Rotation of Presence, the highest peak is Min Presence. And, with regards to those 3 Presence Control settings, the Max position yielded the highest overall impedance, the 50% position showed the widest near-4 ohm impedance range (100Hz-2kHz), then rose up to a broad peak at 25kHz, while the Min position showed the lowest impedance, dropping below 2 ohms @ 6Khz before climbing back up above 4 ohms above 30kHz,

I suppose I could add additional text in Photoshop to better clarify each of the lines in the plot.

I"m about to re-do the Output Impedance plots, if I can raise the fixed load resistor value high enough to have minimum shunt effect on the plotted output impedance. Uncharted territory at the moment, as the amp wouldn't allow any of this under No Load condition.

[nevetslab]

Updated 4 ohm Output Impedance Pllots--raised Load Res to 150 ohms

I found I was able to increase the fixed Resistive Load on the output of the Mesa Simul-Class 290 to 150 ohms without it breaking into LF oscillation or worse. Now, I did find when having the Presence control set to Min & 50%, it WOULD oscillate, but I was able to tame it for that portion of the plot by setting the load to 32 ohms. So those resonance peaks are clipped and loaded down. Once out of the resonance range, I was able to set the load to 150 ohms, which yields much higher output impedance curves, as you'll see below on the previous 4 ohm tap plots and today's efforts.



I also added Measurement notes below, and added further annotation on the graphs to identify the curves and teh resistance calibration lines plotted.

[nevetslab]

Acoustical Measurements-SS Amp vs Tube amp

After completing the Output Impedance Measurements on the Mesa Simul-Class 290 Power Amp, I moved on to the acoustical measurements. The plan was to set the amp output @ 160Hz 1/3 Oct Pink Noise, with the B & K 1027 Generator/1617 Filter and measure the Voltage Output from the power amps @ 5V RMS on the B & K 2607 Measuring Amp. Initially I had the B & K 4134 1/2” Pressure Mic/2619 Preamp @ 1 meter on a short tripod, and made a few measurements using both the Mesa amp and the Alesis RA-100 SS amp. Not being happy with the inconsistency of the 1/6 Oct plot, even with the Averaging on the 2035 Signal Analyzer, I brought in my GenRad 1995 Integrating 1/3 Oct RTA, which can produce more stable readings using the time integration function on the measurement spectrum.

I also injected the 1027 Generator’s Compressor Input from it's ouput for a more stable reading to minimize the min/max noise excursions. I also changed to Ground Plane Measurement mode, with the mics @ 2 meters, directly on the floor. That gave the smoothest readings of the speaker and the two amps. I had previously taken the signal from the 2607 Measuring Amp's output, which was fed from the power amps and injected THAT into the compressor input, which in subsequent plots nullified the Presence & Filter's frequency response changes..







The plots differ a bit, but hold the same spectral trend & the overall SPL Wideband readings were less than 1dB different between the two amps while the GenRad analyzer was reading a bit higher than the B & K. When I began this series of measurements, I had wondered if there would be any significant output level difference between driving the same speaker from a solid state amp vs a tube amp (the Mesa amp in this instance) at the same signal level. That was set at 160Hz 1/3 Oct Pink Noise, at which the speaker is at 'nominal' 4 ohm impedance. The differences in the 1/3 Oct and 1/6 octave pink noise spectrum recorded do show differences, but, while making the measurements, you see all the frequency bands dancing up and down from the randomness of Pink Noise. The B & K Analyzer uses an averaging routine to help smooth that out. The GenRad analyzer instead uses a time-selectable integration of the reading, which does a much better job at achieving a stable non-dancing reading across it's spectrum., and internally is built with discrete filters and integrators, all working in parallel. The B & K readings show the SS amp having a bit higher wide-band SPL reading, while the GenRad's wideband reading shows the Tube amp being a bit higher...in both cases, less than 1dB difference. On the GenRad analyzer, you can simultaneously display Max and Real Time levels, where the Max levels per band show that randomness in the noise. I couldn't capture it on the camera though.

At any rate, the I didn't see any conclusive evidence that there was an SPL difference between the two amp types.

I had set the Presence Control on the Mesa Tube Amp to 50%, which has the widest 4-ohm source impedance over it’s frequency range (100Hz-4kHz). While I haven’t yet measured the source impedance of the Alesis RA-100, I’d guess it to be less than 50 milliohms. I will still check output at the other two Presence Control settings, but won't have the space in this post to show the results.

[drewl]

OOOOhhhhhh Bruell and Kjaar test equipment!

Very fancy! I remember working on that stuff back in the 80's!

[nevetslab]

Ah, YES! It is a real pleasure to work with them. The 2035 2-Ch analyzer that I used (along with the GenRaD 1995) is still a bit of a puzzle. It's a DOS-based machine, and I still don't have the multiple- book-set Operations Manual for it. This box lacks the Generator module, so it's Analyzer only until I can find/afford that precious generator module.that. Calibrating the mic you're using with it is not so straight forward. Not like a simple screw-driver trim pot to set the CAL using a Pistonphone or other style calibrator that fits onto the front of the mic. I'm in the ballpark with it from tweaking the numerical sensitivity values into the system. And, it's NOT a portable instrument by any stretch, weighing over 80 lbs. I know there's a way to set the Averaging time & type, but I have to go back to searching thru the menus and HELP directory. It will drive my graphics plotter, but that's in my storage locker and my fresh pen supply is needing funds. So, I resort to a cheap camera to try and capture the screen image.

Their 1027 Sine-Random Generator....absolutely a wonderful instrument! I use it together with the Type 1617 Filter atop it for generating 1/3 Oct Pink Noise for a lot of things. For instance, setting the filter to 40, 50 or 63Hz, the resultant waveform is a random amplitude sine wave, often having a min-max difference of as much as 15dB. Great for testing powered subs, power amps current limiter circuits when driving a woofer, watching how an amp clips, and great for burning in multi-tiered power amps under acoustic drive, so you're not just pouring steady-state sine into them. So many of the big power amps will NOT produce high output steady state, but under transient signal typical in music, you see the full range. It's a 3-decade oscillator...all with one spin of the dial. Decent linearity...harmonics lower than -80dB. and, like all of their generators, it has the compressor circuit for regulating the output in systems measurments, as I did in this series of measurements...though there I was using the stand-alone compressor in the 1405 Noise Generator to regulate the generator in that USB scope/generator box.

Love the back-lighted meters and dials. Very classy gear, and a real pleasure to work with, as you no doubt remember.