Perhaps it was because these "older" amps had squat for HF response out of the audio band, so hanging a load on the end that looks like a short at a frequency where the amps output drops like a stone has nil effect?
dunno?
_-_-bear
dunno?
_-_-bear
My first comp differential amp that I designed in 1968 did not have (or apparently need) an output coil to be completely stable with any capacitive load. It had about 40 dB of negative feedback and was direct coupled at the input and output. It used complementary output devices with an F(t) of 4 MHz, and the I(q) was 1/2A. I think that the combination of high I(q) and the excessive Miller compensation that I used, kept the amp very stable with load, but it had a fairly lousy slew rate, because of the excessive compensation. (Who cared about slew rate in those days/) 
john curl said:
Hi John,
Now this is getting to be fun!
OK, here is what I saw, right off of the JA measurements (to the best of my eyeballs):
20 kHz: - 0.12 dB at 8 ohms; -0.35 dB at 2 ohms
50 kHz: -1.00 dB at 8 ohms; -1.55 dB at 2 ohms
60 kHz: -1.30 dB at 8 ohms; -2.10 dB at 2 ohms
70 kHz: -1.75 dB at 8 ohms; -2.70 dB at 2 ohms
80 kHz: -2.20 dB at 8 ohms; -3.40 dB at 2 ohms
So, it looks to me like there IS a significant difference between the 2 ohm loading and the 8 ohm loading at 50 kHz and above.
Note that at 70 kHz we are down about 1 dB more at 2 ohms than at 8 ohms. In rough terms, if we assume a pure inductive output, this would correspond to a corner due to the inductance against 2 ohms an octave above, at 140 kHz. The inductance corresponding to such a pole comes out to be 2.3 uH.
I then put together a SPICE frequency response run with a voltage source feeding a 47 ohm resistor shunted to ground by a 0.033 uF capacitor. This would represent the unloaded intrinsic amplifier HF rolloff.
This signal was buffered by a voltage-controlled voltage source with a gain of unity.
That second voltage source then fed a load resistance of either 8 ohms or 2 ohms through a 2.2 uH inductor.
The frequency responses with this setup matched the readings I got from the JA curves to within about 0.2 dB. A pretty good fit. This tells me that even without a coil, JA's measurements suggest that the effective output inductance of the JC-1 is about 2.2 uH.
So, I'm not sure you've got me 🙂.
Let me know what you think. If I've screwed up here or made some wrong assumptions, it would not be the first time. I'm just reporting my interpretation of the data that I saw, and agree that such inferences need to be taken cautiously.
Cheers,
Bob
My amp is smarter than yours! It has active output protection! And there is a difference between I(q) and IQ.
Hi Bob,
so from the number of 8 ohm and 2 ohm loading..........
does this mean that people don't use coil so that the amp frequency response doesn't change much with differing loading ?
Hartono.
so from the number of 8 ohm and 2 ohm loading..........
does this mean that people don't use coil so that the amp frequency response doesn't change much with differing loading ?
Hartono.
forr said:
Let me see if I can be more clear about what I said in regard to transmission line effects in speaker cables.
Light travels at about 6 inches per ns in most cables. Let's take a 10-foot speaker cable. The round trip delay will be about 40 ns. I've actually measured several by a pulse method and usually get about 50 ns. So we are in the ballpark.
For transmission lines, one of the magic numbers is the quarter-wave length of the cable. At this point, the round trip is a half-wave, or 180 degrees. At this frequency, an ideal unterminated transmission line will look like a short at its input, since whatever goes in will be reflected from the far end and arrive back at the input 180 degrees out of phase.
A quarter wave at 10 MHz is 25 ns. This means that at 10 Mhz, with a 10-foot speaker cable exhibiting a round-trip delay of 50 ns, we have the quarter-wave condition.
10 MHz is clearly above the gain crossover frequency of most audio amplifiers, but the capacitive slope of the unterminated line will start to create effects at lower frequencies, perhaps only an octave or two above an amplifier with a gain crossover of 2 MHz.
Moreover, consider local output stage stability effects - not just global feedback effects. If we have an EF output stage with 30 mhz or 60 Mhz ft output devices driving an unterminated line that is a quarter wave at 10 Mhz, the potential for bad things happening is definitely there. EF's not isolated by resistance and/or inductance often don't like driving wierd loads.
So this is what I meant by transmission line effects.
BUT, you might say (as I often have), most speakers will look resistive at high frequencies because they must pad down the tweeter. That's what I thought. All of my speakers pad down the tweeter so at HF they look pretty much like 8 ohms resistive in spite of any tweeter voice coil inductance.
WRONG!
Just look at the review of the new Wilson WATT/Puppy 8 in Stereophile. From the impedance curve, the impedance is about 20 ohms at 50 kHz and still rising! Looks like Dave Wilson is driving the tweeter wide open! Probably so, since his voltage sensitivity for that speaker is about 92 dB. No pad needed.
So this speaker leaves the speaker cable essentially unterminated at high frequencies. This speaker would be an ideal candidate for a Zobel network at the speaker input terminals. That would terminate the speaker cable at high frequencies and destroy the quarter-wave reflection at 10 Mhz.
Cheers,
Bob
jacco vermeulen said:
Hey pal.
I have finally found out what you mean by above post. Yes, it took me one day. Why so long? Because I never expected such totally ridiculous indictment, or at least insinuation of violating rule #5.
I do hope you realize you are violating rule #1. If not, maybe the moderators will help you.
Then you publicly accused me of 'lacks every bit of integrity'.
Why? Because I once lost my patience with only one person, who's nota bene banned from this forum?
Or don't you like my jokes about class-heat, sorry, class-A?
Again, it's you who's violating rule #1, speaking of integrity!
Besides, suppose it's true what you're saying, why I'm never sin-binned for at least a couple of days?
john curl said:My amp is smarter than yours! It has active output protection! And there is a difference between I(q) and IQ.
I'm confused, John. What does this have to do with the small-signal output frequency response into 8 ohms and 2 ohms. I assume your protection circuitry is not coming into play at a mere 4 watts into 2 ohms.
Cheers,
Bob
Be careful with Bob's graphical interpretation of the JC-1. You HAVE to match the curves at low frequencies first, in order to make a fair assessment of any frequency change of the amp with loading. The primary rolloff of the high frequency response is at the input of the amp and perhaps the lead cap across the feedback resistor. My calculated derived output inductance is about 0.125uH due the the rate of change of the output impedance due to the finite open loop bandwidth of a few KHz and the very low open loop output impedance.
This also is inferred by 'Stereophile's' measurements, but there is no engineering reason that it would be more than that. It 'might' be as much as 0.2uH worst case, if I needed to worst case, but that is still better than a factor of 10 over what Bob predicts.
It is also effectively half as much as Halcro, in any case.
This also is inferred by 'Stereophile's' measurements, but there is no engineering reason that it would be more than that. It 'might' be as much as 0.2uH worst case, if I needed to worst case, but that is still better than a factor of 10 over what Bob predicts.
It is also effectively half as much as Halcro, in any case.
john curl said:My first comp differential amp that I designed in 1968 did not have (or apparently need) an output coil to be completely stable with any capacitive load. It had about 40 dB of negative feedback and was direct coupled at the input and output. It used complementary output devices with an F(t) of 4 MHz, and the I(q) was 1/2A. I think that the combination of high I(q) and the excessive Miller compensation that I used, kept the amp very stable with load, but it had a fairly lousy slew rate, because of the excessive compensation. (Who cared about slew rate in those days/)
Good point, John.
Bob
bear said:So, now John Curl is claiming an amp with IQ??
_-_-bear
However, an IQ of 0.5 isn't particular high. 😉
Well, at RF :
- big crossover caps are RLC resonant circuits because of the inductance
- big inductors are RLC resonant circuits because of the parasitic capacitance
- I doubt power wirewound resistors have any resistive behaviour left in them
So basically, I wonder if someone dared to connect a network analyzer to a speaker cable, at the end of which is connected a speaker, and plot the impedance and return loss from DC to RF...
- big crossover caps are RLC resonant circuits because of the inductance
- big inductors are RLC resonant circuits because of the parasitic capacitance
- I doubt power wirewound resistors have any resistive behaviour left in them
So basically, I wonder if someone dared to connect a network analyzer to a speaker cable, at the end of which is connected a speaker, and plot the impedance and return loss from DC to RF...
Hi John,
I think most people here wouldn't care if the response change with different loading. Is this the only reason for not using output coil ?
Hartono
I think most people here wouldn't care if the response change with different loading. Is this the only reason for not using output coil ?
Hartono
john curl said:
John,
The differences between the 8 ohm loading and the 2 ohm loading at low frequencies is microscopic for the JC-1, so your comment is a red herring.
I really get a laugh out of you, of all people, saying that something cannot be because there is no engineering reason for it!!!!!
GOTCHA!!
Cheers,
Bob
john curl said:My first comp differential amp that I designed in 1968 did not have (or apparently need) an output coil to be completely stable with any capacitive load. It had about 40 dB of negative feedback and was direct coupled at the input and output. It used complementary output devices with an F(t) of 4 MHz, and the I(q) was 1/2A. I think that the combination of high I(q) and the excessive Miller compensation that I used, kept the amp very stable with load, but it had a fairly lousy slew rate, because of the excessive compensation. (Who cared about slew rate in those days/)
Please Mr Curl, could you be a little more specific: 40dB of NFB at what frequency?
And how large was your excessive Miller compensation?
I do like to learn from you, and, of course from exact numbers. 🙂
Cheers,
peufeu said:
I did, and replied in the thread linked by Forr (RFI). Probably nobody seen, as everyone wants to medialize here 😀
peufeu said:
peufeu said:
So, peufeu, what are you trying to say? 😀
Doesn’t the signal need to be large enough to physically engage the drivers so the back-EMF starts to play a role? Also, shouldn’t the signal be impulse or burst based?
PMA: Looooong thread. Which post was your reply in?
- Status
- Not open for further replies.