Heatsinking Requirements
The test results for heatsinking requirements are now up at http://home.pacific.net.au/~gnb/audio/lm4702.html#heatsinking.
Does the LM4702 need a heatsink? Yes and No.
No, up to about +/- 45 V on the power supply rails. Yes, at least a small heatsink should be used above +/- 45 V.
Next up is some interesting information on the GND pin.
glennb said:
The test results for heatsinking requirements are now up at http://home.pacific.net.au/~gnb/audio/lm4702.html#heatsinking.
Does the LM4702 need a heatsink? Yes and No.
No, up to about +/- 45 V on the power supply rails. Yes, at least a small heatsink should be used above +/- 45 V.
Next up is some interesting information on the GND pin.
Re: Heatsinking Requirements
Nice work. Might be helpful if you put some of the graphical data in Excel. (for all those waiting -- Google is beta-testing a web-spreadsheet program -- you need a Google account -- easy to set up -- here's a link http://spreadsheets.google.com ).
For the moment I am limited in my testing with my +/- 34V Trygon Supply (it will do 20 amps) -- so the initial tests will probably be ~120 watts or so. I am loathe to move around a pair of the HP6129's as they weigh in at ~60 pounds each.
I would mention on your site that National took pains to shield the torroids with mu-metal -- every little bit helps when you are tring to get the last oink out of the pig.
glennb said:
The test results for heatsinking requirements are now up at http://home.pacific.net.au/~gnb/audio/lm4702.html#heatsinking.
Does the LM4702 need a heatsink? Yes and No.
No, up to about +/- 45 V on the power supply rails. Yes, at least a small heatsink should be used above +/- 45 V.
Next up is some interesting information on the GND pin.
Nice work. Might be helpful if you put some of the graphical data in Excel. (for all those waiting -- Google is beta-testing a web-spreadsheet program -- you need a Google account -- easy to set up -- here's a link http://spreadsheets.google.com ).
For the moment I am limited in my testing with my +/- 34V Trygon Supply (it will do 20 amps) -- so the initial tests will probably be ~120 watts or so. I am loathe to move around a pair of the HP6129's as they weigh in at ~60 pounds each.
I would mention on your site that National took pains to shield the torroids with mu-metal -- every little bit helps when you are tring to get the last oink out of the pig.
Re: Re: Heatsinking Requirements
OK, I've mentioned the mu-metal shielding used in the AN-1490 amplifier. The Excel spreadsheet which generated the temperature graph is now available at http://home.pacific.net.au/~gnb/audio/lm4702.html#heatsinking . I'm currently putting together the GND pin info and will post here when its up on the web page.
jackinnj said:
OK, I've mentioned the mu-metal shielding used in the AN-1490 amplifier. The Excel spreadsheet which generated the temperature graph is now available at http://home.pacific.net.au/~gnb/audio/lm4702.html#heatsinking . I'm currently putting together the GND pin info and will post here when its up on the web page.
Google Robot
Hi,
Since Google Bot is using my words for the header to this thread,
I feel reponsible to fix the ambiguity.
I matched two (2) Application Notes:
www.national.com/ds/LM/LM3886.pdf
www.National.com/an/AN/AN-1192.pdf
The 3886 note shows at page 15, IMD 60 Hz, 7 kHz 1:1 (No
Figure Number) a normal distribution around a 68 to 72 Hz line
signal.
This 3886 note was compared with 1192 nosie floor notes at
Figure 11 wherein National said "an FFT analyzer is extremely
handy in determining the noise culprit" and Figure 12 "even with
the limited number of graphs showen, the quality of this amplifier
[LM4702] from a measurement perspective is quite good".
So all I ment to say, before Google Bot caught the words in mix-
match, was that App Note 3886 has more graphs than App Note
1192, since the LM-4702 FFT noise floor tests in THD, were done
with or without an output stage in the equation (question?)
where the LM-3886 tests were done with integrated output
stage from 60Hz to 80Hz on page 15 of same App Note.
Is the Mu metal for evil 60Hz signals that go within reasonable
harmonics or is the Mu metal to control higher frequency signals
and harmonics?
Sorry for the Mix-up on Google Bot.
Mark
Hi,
Since Google Bot is using my words for the header to this thread,
I feel reponsible to fix the ambiguity.
I matched two (2) Application Notes:
www.national.com/ds/LM/LM3886.pdf
www.National.com/an/AN/AN-1192.pdf
The 3886 note shows at page 15, IMD 60 Hz, 7 kHz 1:1 (No
Figure Number) a normal distribution around a 68 to 72 Hz line
signal.
This 3886 note was compared with 1192 nosie floor notes at
Figure 11 wherein National said "an FFT analyzer is extremely
handy in determining the noise culprit" and Figure 12 "even with
the limited number of graphs showen, the quality of this amplifier
[LM4702] from a measurement perspective is quite good".
So all I ment to say, before Google Bot caught the words in mix-
match, was that App Note 3886 has more graphs than App Note
1192, since the LM-4702 FFT noise floor tests in THD, were done
with or without an output stage in the equation (question?)
where the LM-3886 tests were done with integrated output
stage from 60Hz to 80Hz on page 15 of same App Note.
Is the Mu metal for evil 60Hz signals that go within reasonable
harmonics or is the Mu metal to control higher frequency signals
and harmonics?
Sorry for the Mix-up on Google Bot.
Mark
Mu Metal
Hi,
Mu Metal for beginners like myself:
http://digilander.libero.it/paeng/what_mu_metal.htm
which is about .77 Nickel and .15 Iron and trace copper and trace
molybenum.
Mu Metal (TM) is registered trade mark of Telcom Metals Ltd.
Mark
Hi,
Mu Metal for beginners like myself:
http://digilander.libero.it/paeng/what_mu_metal.htm
which is about .77 Nickel and .15 Iron and trace copper and trace
molybenum.
Mu Metal (TM) is registered trade mark of Telcom Metals Ltd.
Mark
XDA-1 Output Stages
Hi,
Check out the new output stages on the XDA-1.
Things besides and including 1-2-3 Darlington pairs to get to the
top.
Oh, the same old output stages? No way. New ideas. Maybe a
complete review of all superior output stages. This designer goes
for the max. Why limit output stages to 1-2-3 Darlington? Are
they they best? If there is a so called "best" or just subjective
flavors.
Peace
Mark
Hi,
Check out the new output stages on the XDA-1.
Things besides and including 1-2-3 Darlington pairs to get to the
top.
Oh, the same old output stages? No way. New ideas. Maybe a
complete review of all superior output stages. This designer goes
for the max. Why limit output stages to 1-2-3 Darlington? Are
they they best? If there is a so called "best" or just subjective
flavors.
Peace
Mark
Why Serrial And Not Parallel Outout Stages
Hi,
They do not want my redundancy in this URL so
here is http://genzbenz.com/? -- if you would like
to, please go find the parallell output stages FET and 12AX7 tube
stages as cited in a previous post on this thread.
Mark
Hi,
They do not want my redundancy in this URL so
here is http://genzbenz.com/? -- if you would like
to, please go find the parallell output stages FET and 12AX7 tube
stages as cited in a previous post on this thread.
Mark
Mute current
Yes, I did a quick experiment with lowering the mute current. The test was at Vcc/Vee=+/-40V and driving 1KHz sine at 10Vpeak into a 3K load, which is close to the maximum output drive current.
The +ve peak of the output waveform becomes clipped when Imute is below 0.36mA. Clipping is at +6V at Imute=0.27mA.
The moral is to follow NS's guidance to keep the mute current above 1mA to ensure the chip is completely in "play" mode, but don't go above the 2mA maximum spec.
jackinnj said:
Yes, I did a quick experiment with lowering the mute current. The test was at Vcc/Vee=+/-40V and driving 1KHz sine at 10Vpeak into a 3K load, which is close to the maximum output drive current.
The +ve peak of the output waveform becomes clipped when Imute is below 0.36mA. Clipping is at +6V at Imute=0.27mA.
The moral is to follow NS's guidance to keep the mute current above 1mA to ensure the chip is completely in "play" mode, but don't go above the 2mA maximum spec.
GND pin on LM4702
The test results for my initial investigations into the GND pin of the LM4702 are now available at http://home.pacific.net.au/~gnb/audio/lm4702.html#groundpin
In summary:
The GND pin is used as an internal reference point for constant current sources and other signals. It draws between 1.7 and 1.8 milliAmps towards the negative Vee power supply rail. When amplifying a signal to the maximum output current and voltage capability the GND pin also carries a 0.3 milliAmp peak-to-peak modulation current which is distorted version of the output voltage; the peaks of the waveform are noisy and flattened.
I recommend that GND pin should be directly connected to the "star earth" point as used by the central 0V point of power supply, the speaker output common, the Zobel network, the power supply bypass capacitors, and the low level "clean earth" point. The "clean earth" is used for the input signal and feedback network, and is usually isolated by a thin PCB track or a low value resistor from the large currents which circulate through the "star earth" point.
The GND pin should not be connected into the path of the "clean earth" as it will produce unwanted DC and AC interaction with this particularly sensitive part of the circuit.
As always, I am open to discussion and differing opinions.
The test results for my initial investigations into the GND pin of the LM4702 are now available at http://home.pacific.net.au/~gnb/audio/lm4702.html#groundpin
In summary:
The GND pin is used as an internal reference point for constant current sources and other signals. It draws between 1.7 and 1.8 milliAmps towards the negative Vee power supply rail. When amplifying a signal to the maximum output current and voltage capability the GND pin also carries a 0.3 milliAmp peak-to-peak modulation current which is distorted version of the output voltage; the peaks of the waveform are noisy and flattened.
I recommend that GND pin should be directly connected to the "star earth" point as used by the central 0V point of power supply, the speaker output common, the Zobel network, the power supply bypass capacitors, and the low level "clean earth" point. The "clean earth" is used for the input signal and feedback network, and is usually isolated by a thin PCB track or a low value resistor from the large currents which circulate through the "star earth" point.
The GND pin should not be connected into the path of the "clean earth" as it will produce unwanted DC and AC interaction with this particularly sensitive part of the circuit.
As always, I am open to discussion and differing opinions.
I have run the amp up to 100 W with single set of Sanken devices -- and it was stable and cookin' -- but the output is limited to what my big bench supply can provide. Some nice big 39,000uF caps just showed up from DK so expect more news in August.
To get the best performance numbers you have to be very, very careful with the layout so spend your time looking at the layout in the NatSemi apnote. The first time I ran the amp I had a 5.6MHz (Megahertz) oscillation even with belt and suspenders output LP filtering.
As I said earlier, I am trying to get my Wide-Beveled Thors up and running so the amp is still sitting in the lab table.
To get the best performance numbers you have to be very, very careful with the layout so spend your time looking at the layout in the NatSemi apnote. The first time I ran the amp I had a 5.6MHz (Megahertz) oscillation even with belt and suspenders output LP filtering.
As I said earlier, I am trying to get my Wide-Beveled Thors up and running so the amp is still sitting in the lab table.
LM4702B
Yes, the PDF data sheet has been revised for the LM4702B, which now has uprated power supply rails of +/- 100V, as against +/- 80V in the pre-announcement last year. It doesn't look like samples are available yet.
The previoulsly released LM4702C which people have been experimenting with has rail limits of +/- 75V.
The LM4702A is still under development. It will have the same rail limits as the LM4702B (+/- 100V) but in a TO-3 military compliant package, as against +/- 85V in the pre-announcement last year.
1000 Wrms into 4 Ohm anyone?
paulb said:
Yes, the PDF data sheet has been revised for the LM4702B, which now has uprated power supply rails of +/- 100V, as against +/- 80V in the pre-announcement last year. It doesn't look like samples are available yet.
The previoulsly released LM4702C which people have been experimenting with has rail limits of +/- 75V.
The LM4702A is still under development. It will have the same rail limits as the LM4702B (+/- 100V) but in a TO-3 military compliant package, as against +/- 85V in the pre-announcement last year.
1000 Wrms into 4 Ohm anyone?
- Status
- This old topic is closed. If you want to reopen this topic, contact a moderator using the "Report Post" button.