Is this a legitimate approach: inject an audio signal at the input using a cheapo BK audio signal generator. With my Fluke DMM, measure AC voltage at the input to the amp and at the output, with a speaker attached. Compare the voltage "gain" at various frequencies.

With my Scott All-Wave 15 radio, c. 1934, using the phono connection as the input, I get these results: between 200 and 15,000 Hz, response is within plus/minus 2 dB. Good, but not tremendous.

Below 200 Hz, the response drops off quickly, down 4 dB between 200 and 100 Hz.

Above 15,000 Hz, things don't seem to be rolling off too quickly. 20,000 Hz is only 2 dB below 15,000 Hz.

The reasonable high-frequency response indicates to me that the audio interstage xfmrs are pretty good. I'm not sure how to explain the poor low-freq response. Remember, this is for the amp only, not the speaker.

Interstage transformers will tend to be less efficient at lower frequencies, like any transformer. A 50 cycle transformer, for instance, will work fine at 60 cycles, but reversing this will reduce the efficiency of the transformer (it won't be anything dramatic like trying to operate either at 25 cycles).

With higher frequencies, high impedance may limit these, if the transformer has enough windings (which will also aid low frequency response), but capacitive coupling, if the windings are right on top of eachother, may help null this problem. Also, capacitive coupling in each winding (not between the windings), may actually reduce high frequency response by shunting it across the coil instead of through it.

Something I've thought about a lot is that when people take these measurements, they never take into consideration the frequency response of the measuring devices, or the supposedly 'constant' devices. I wonder how much affect this has on things. Certainly an AC meter will respond fairly well to most frequencies, so I guess that this is not a problem here. Considering, though, that everything is relative, things can be a bit more complicated than they seem.

Thomas

Thomas: My Fluke 179 is supposed to be good up to 50 kHz. But that spec is for measuring frequency, not AC voltage. I guess we can assume that there wouldn't be a problem?

The problem I noticed was reading the low frequencies, e.g. <200 Hz. The DMM's display refreshes 4 times per second, so the display keeps changing numbers as the voltage goes through a cycle. Maybe I should rerun the low-freq portion of my test with my Simpson meter that is mechanical/analog.

By the way, after running that test, my ears are ringing. I used one of my hi-fi speakers as the load, and the high-pitched tones were piercing. Once I got up to 12 kHz and above, probably every dog in the neighborhood covered his ears. I didn't want to adjust the volume after starting the test to keep everything on the level.

You should take a crappy speaker and rip out its cone. Just leave the voice coil in place so that it has proper iron around it.

T.

Good idea. I'll remember that next time.

When I said my ears are still ringing, I'm not kidding.

I'm not an audiologist, but my testing showed that anything above 12 kHz, I'm totally deaf. (Well, at least I am now after the testing!) Up above that frequency, I have to rely on a voltmeter and/or a dog.

:You should take a crappy speaker and rip out its cone. Just leave the voice coil in place so that it has proper iron around it.
:
:T.

Hi Doug,
Here's something to consider as to why low frequency response drops off quickly:
Early power supplies were not very efficient in filtering the 120Hz ripple from a full-wave rectifier. As a result, the plate decoupling (RC) circuits in many radios were designed to filter out frequencies below about 130Hz. This took care of the 120Hz ripple frequency getting through, but low frequency response also suffered.

Put on you favorite records that are mostly below 15khz tones...
....then if you can just get the wife to speak at 15khz or above ....you will have total audio bliss

Doug, Did the ringing in your ears go away yet ? I ended up with tinnitus for 3 days after doing pretty much the same thing.

Doug,
What does the speaker impedance actually look like at 200Hz? Keep in mind that all the impedances in the string must maintain in order to get a predictable response curve. If the interstage/output transformers are after market newer assemblies, the core material may well be of a better mixture than that of yesteryear, providing somewhat enhanced performance. Still, I'd say that your findings are pretty respectable for a set that old. In addition, your loss of hearing may well be a blessing as you motor down the by-ways, ending up beside a rice burner with a KW stereo. Put on a Glenn Miller, kick back the recliner, and enjoy.

marv

:Is this a legitimate approach: inject an audio signal at the input using a cheapo BK audio signal generator. With my Fluke DMM, measure AC voltage at the input to the amp and at the output, with a speaker attached. Compare the voltage "gain" at various frequencies.
:
:With my Scott All-Wave 15 radio, c. 1934, using the phono connection as the input, I get these results: between 200 and 15,000 Hz, response is within plus/minus 2 dB. Good, but not tremendous.
:
:Below 200 Hz, the response drops off quickly, down 4 dB between 200 and 100 Hz.
:
:Above 15,000 Hz, things don't seem to be rolling off too quickly. 20,000 Hz is only 2 dB below 15,000 Hz.
:
:The reasonable high-frequency response indicates to me that the audio interstage xfmrs are pretty good. I'm not sure how to explain the poor low-freq response. Remember, this is for the amp only, not the speaker.

Try your test with a resistive load. The impeadence of the speaker is going to change with the frequency so the loading on the circuit will change. The speaker is just a noisy inductor. The speaker will give you skewed results. Let us curious ones know if there is different results.

D. Wit

:Is this a legitimate approach: inject an audio signal at the input using a cheapo BK audio signal generator. With my Fluke DMM, measure AC voltage at the input to the amp and at the output, with a speaker attached. Compare the voltage "gain" at various frequencies.
:
:With my Scott All-Wave 15 radio, c. 1934, using the phono connection as the input, I get these results: between 200 and 15,000 Hz, response is within plus/minus 2 dB. Good, but not tremendous.
:
:Below 200 Hz, the response drops off quickly, down 4 dB between 200 and 100 Hz.
:
:Above 15,000 Hz, things don't seem to be rolling off too quickly. 20,000 Hz is only 2 dB below 15,000 Hz.
:
:The reasonable high-frequency response indicates to me that the audio interstage xfmrs are pretty good. I'm not sure how to explain the poor low-freq response. Remember, this is for the amp only, not the speaker.

:D.Witt told you right. Never run an amp freq response with the speaker as a load as it represents an erratic inductor with constatly changing Z. You need a NON REACTIVE resistance load. I recommend you use either carbon or metal film --NOT WIRE WOUND. I personally use a bank of carbon resistors suspended in a one quart paint can filled with silicon oil enabling me to dissipate 400 watts or so. Good luck.

Try your test with a resistive load. The impeadence of the speaker is going to change with the frequency so the loading on the circuit will change. The speaker is just a noisy inductor. The speaker will give you skewed results. Let us curious ones know if there is different results.
:
:D. Wit
:
:
:
::Is this a legitimate approach: inject an audio signal at the input using a cheapo BK audio signal generator. With my Fluke DMM, measure AC voltage at the input to the amp and at the output, with a speaker attached. Compare the voltage "gain" at various frequencies.
::
::With my Scott All-Wave 15 radio, c. 1934, using the phono connection as the input, I get these results: between 200 and 15,000 Hz, response is within plus/minus 2 dB. Good, but not tremendous.
::
::Below 200 Hz, the response drops off quickly, down 4 dB between 200 and 100 Hz.
::
::Above 15,000 Hz, things don't seem to be rolling off too quickly. 20,000 Hz is only 2 dB below 15,000 Hz.
::
::The reasonable high-frequency response indicates to me that the audio interstage xfmrs are pretty good. I'm not sure how to explain the poor low-freq response. Remember, this is for the amp only, not the speaker.

A 400-W resistor bank to substitute for the speaker might be a little overkill for this radio? That's about four times the power that the whole radio draws from the AC line. My pair of 2A3 audio output tubes, in push-pull, puts out about 10W, max.

::D.Witt told you right. Never run an amp freq response with the speaker as a load as it represents an erratic inductor with constatly changing Z. You need a NON REACTIVE resistance load. I recommend you use either carbon or metal film --NOT WIRE WOUND. I personally use a bank of carbon resistors suspended in a one quart paint can filled with silicon oil enabling me to dissipate 400 watts or so. Good luck.
:

Why only 10 watts? I can get that from one 2A3 (internally parallel connected). Also, would a bucket of water be as effective as silicon oil? Silicon brake fluid is rather expensive.

Also, regarding power supplies, actually, the more the power supply filters out low frequency passage, the better your low frequency response will be. You see, if the power supply is flimsy at low frequencies, and allows them to pass through *IT*, it will be flimsy with the low frequencies that the amplifier is trying to put out. That is, the loading of a low frequency will be able to load down the voltage of the power supply for each low frequency wave. If a power supply is designed to really filter out all low frequency passage through IT (notice how I emphasize IT, as in, the power supply), it'll really hold steady for each low frequency wave fluctuation of the amplifier. Old fashioned power supplies do have low frequency filters, i.e. chokes, but are actually weaker than modern power supplies at filtering out low frequencies. Modern power supplies use large condensers with brute force to filter out low frequency fluctuations. The more the power supply can hold a constant voltage for each shift in load for each wave the amplifier amplifies, the better your bass response will be. If the power supply loads down and reduces voltage output for each up swing in current draw by the amplifier for each wave of any particular low frequency, it'll work like negative feedback, and will cancel what the amplifier is trying to put out.

T.

T.

To clarify more what I said, a power supply filters out low frequency passage through it by filling in the troughs of each wave of the rippling DC voltage passing through it. If it is flimsy at this attempt, then it will be less likely to hold a steady voltage for any low frequency load shifts imposed upon it.

If you want solid bass out of your amplifier, make sure that the bass is filtered out completely from your power supply. The power supply must be able to maintain as pure and as constant a voltage as possible.

You may get an effect of more bassiness with a power supply that is dying or weak (filters going bad or of small capacity). This is because the power supply is being loaded down by each low frequency ripple, which may, under some circumstances, accentuate bass (not sure how, because it should actually cancel out), but you will only get solid bass notes at higher volume levels if the power supply is beefy and strong. I say this because I have had amplifiers with weak filteres that seemed more bassy at low volume levels than they were with the filters replaced (and increased in size), but the amplifiers always performed better at high volumes with new filters. Perhaps the previous effect was in my imagination, because usually bad filter condensers ruin bass response for obvious reasons. One thing that is certain is that weak filters will allow audio in the output to get into the input, which may have been what I was hearing. Any audio fluctuations in the power supply due to bad filters and heavy loads from the audio output, will transfer to other stages within the radio or amplifier. This can even create feedback. My Crosley 1117, for instance, would buzz with a very low frequency feedback when really strong output tubes were inserted, and had really bassy music, but no power. Beefing up the filteres eliminated this problem.

Thomas

Thomas, I'm just going by the RCA Receiving Tube Manual (RC-30), plus the tube data here on Nostalgiaair. For a single 2A3, it lists the output power as 3.5 W.

For push-pull, it has two lines of data, one showing 10W output and the other 15W. 10W seems to apply to my setup - 5K-ohm plate load. For a 3K-ohm plate load, it lists 15W as the output.

The output transformer recommended by Hammond for 2A3s in push-pull is rated at 15W.

I intend to rerun my freq response test with a resistive load, but I'm going to have to order the resistors. In addition the the freq testing, I'll run the set with volume turned all the way up -- and measure the voltage to calculate the output power.

.
:Why only 10 watts? I can get that from one 2A3 (internally parallel connected). Also, would a bucket of water be as effective as silicon oil? Silicon brake fluid is rather expensive.
:
:Also, regarding power supplies, actually, the more the power supply filters out low frequency passage, the better your low frequency response will be. You see, if the power supply is flimsy at low frequencies, and allows them to pass through *IT*, it will be flimsy with the low frequencies that the amplifier is trying to put out. That is, the loading of a low frequency will be able to load down the voltage of the power supply for each low frequency wave. If a power supply is designed to really filter out all low frequency passage through IT (notice how I emphasize IT, as in, the power supply), it'll really hold steady for each low frequency wave fluctuation of the amplifier. Old fashioned power supplies do have low frequency filters, i.e. chokes, but are actually weaker than modern power supplies at filtering out low frequencies. Modern power supplies use large condensers with brute force to filter out low frequency fluctuations. The more the power supply can hold a constant voltage for each shift in load for each wave the amplifier amplifies, the better your bass response will be. If the power supply loads down and reduces voltage output for each up swing in current draw by the amplifier for each wave of any particular low frequency, it'll work like negative feedback, and will cancel what the amplifier is trying to put out.
:
:T.
:
:T.

Ah, well, if you have a dual plate 2A3, the output should be more. Single plate 2A3s are what put out 3.5 watts. A dual plate 2A3, I am guessing, is the same as running two single plate 2A3s in parallel. I don't know how these two tubes can even be sold under the same number, but somehow they are. I never understood this. If you have any single plate 2A3s and double plate 2A3s, compare them and let me know. I only have the double kind, and I'm not sure what all the rave is about the single plate ones on eBay. The double plate one certainly gives you more bang for your buck.

Also, your interstage transformers may be what's cutting off the low frequency response, because the 2A3 can really move a speaker. It's impressive! I need to get my amplifier going again. I was really satisfied with it.

T.

Doug, I don't think anyone picked up on the water in the paint can as opposed to silicon. Don't use water! The resistors and the connections are not water tight and you will end up shorting something out or injuring yourself. The silicon will not conduct but unless you have pure water, water will conduct. The siliconis there to disipate the heat from the resistors, but your not up to 400 Watts to disipate and could use a fan on the resistors to aid in colling. PL

Well, if you had the resistor on the secondary of the output transformer, wouldn't it kind of be isolated from the B circuit? It also wouldn't have more than a maximum of 6 volts on it with a typical amplifier no more powerful than 10 watts. I was able to light a 6L6G on the output of a Zenith consol from the 1950s with two 6V6GTs. That consol had terriffic sound with the Cobra phonograph and all. It's amazing what Zenith could put into their radios. The AM was almost as static free and good sounding as the FM!

Thomas

It's not so much the wattage as it is the resistance of the load that determines how much current it will take. The higher the wattage of the load, in this case 400 watts of say serveal resistors in series/parallel to match the output impedance of the driver (i.e., radio), generally means that your load will not heat up as quickly to the point of burning up and becoming a short to the output, which would be a bad thing.

If as someone else indicated in the discussion of using water in the can to cool the load, do not use WATER as it is conductive and will look like a short between the connecting leads to the load.

:A 400-W resistor bank to substitute for the speaker might be a little overkill for this radio? That's about four times the power that the whole radio draws from the AC line. My pair of 2A3 audio output tubes, in push-pull, puts out about 10W, max.
:
:::D.Witt told you right. Never run an amp freq response with the speaker as a load as it represents an erratic inductor with constatly changing Z. You need a NON REACTIVE resistance load. I recommend you use either carbon or metal film --NOT WIRE WOUND. I personally use a bank of carbon resistors suspended in a one quart paint can filled with silicon oil enabling me to dissipate 400 watts or so. Good luck.
::
:

How did the freq response turn out with a resistor load?

Dim

:Is this a legitimate approach: inject an audio signal at the input using a cheapo BK audio signal generator. With my Fluke DMM, measure AC voltage at the input to the amp and at the output, with a speaker attached. Compare the voltage "gain" at various frequencies.
:
:With my Scott All-Wave 15 radio, c. 1934, using the phono connection as the input, I get these results: between 200 and 15,000 Hz, response is within plus/minus 2 dB. Good, but not tremendous.
:
:Below 200 Hz, the response drops off quickly, down 4 dB between 200 and 100 Hz.
:
:Above 15,000 Hz, things don't seem to be rolling off too quickly. 20,000 Hz is only 2 dB below 15,000 Hz.
:
:The reasonable high-frequency response indicates to me that the audio interstage xfmrs are pretty good. I'm not sure how to explain the poor low-freq response. Remember, this is for the amp only, not the speaker.

I've ordered some resistors to make up a load bank. I'll probably rerun the test in a week or so.

:How did the freq response turn out with a resistor load?
:
:Dim
:
:
:
::Is this a legitimate approach: inject an audio signal at the input using a cheapo BK audio signal generator. With my Fluke DMM, measure AC voltage at the input to the amp and at the output, with a speaker attached. Compare the voltage "gain" at various frequencies.
::
::With my Scott All-Wave 15 radio, c. 1934, using the phono connection as the input, I get these results: between 200 and 15,000 Hz, response is within plus/minus 2 dB. Good, but not tremendous.
::
::Below 200 Hz, the response drops off quickly, down 4 dB between 200 and 100 Hz.
::
::Above 15,000 Hz, things don't seem to be rolling off too quickly. 20,000 Hz is only 2 dB below 15,000 Hz.
::
::The reasonable high-frequency response indicates to me that the audio interstage xfmrs are pretty good. I'm not sure how to explain the poor low-freq response. Remember, this is for the amp only, not the speaker.