LM13700 weirdness
in [Design]
|
Prev: SPSC2 Does anybody make a micro with 3 DACs
Next: Ocean County NJ Family-Law and Cyber-Law Lawyer Charles Novins Continues To Defend Against Net Abusers
From: Bitrex on 26 Apr 2010 17:49 Jim Thompson wrote: > On Mon, 26 Apr 2010 06:07:04 -0400, Bitrex > <bitrex(a)de.lete.earthlink.net> wrote: > >> Jim Thompson wrote: >>> On Sun, 25 Apr 2010 19:57:59 -0400, Bitrex >>> <bitrex(a)de.lete.earthlink.net> wrote: >>> >>>> Joerg wrote: >>>>> Bitrex wrote: >>>>>> Phil Hobbs wrote: >>>>>>> On 4/24/2010 6:25 PM, Bitrex wrote: >>>>>>>> Phil Hobbs wrote: >>>>>>>>> On 4/24/2010 2:23 AM, Bitrex wrote: >>>>>>>>>> I'm using an LM13700 as a VCA in a 9V battery powered circuit, >>>>>>>>>> using an >>>>>>>>>> opamp buffered virtual ground. The circuit works fine when powered >>>>>>>>>> off >>>>>>>>>> the battery, but I wanted to test it using different supply voltages. >>>>>>>>>> When I hooked up the supply to the 15 volt breaboard power supply, I >>>>>>>>>> noticed that the LM13700 was oscillating at about 150 Hz when >>>>>>>>>> there was >>>>>>>>>> no input signal - the oscillation stops when signal is applied and >>>>>>>>>> the >>>>>>>>>> VCA turns on. I couldn't seem to get it to settle down with the usual >>>>>>>>>> supply bypass capacitors, whatever values I tried. Maybe it's just a >>>>>>>>>> consequence of sloppy breadboard layout? The other ICs in the circuit >>>>>>>>>> (mostly TL084s) work OK regardless of supply choice. Suggestions >>>>>>>>>> would >>>>>>>>>> be appreciated! >>>>>>>>> Motorboating? >>>>>>>>> >>>>>>>>> Cheers >>>>>>>>> >>>>>>>>> Phil Hobbs >>>>>>>>> >>>>>>>> Maybe, but I've tried decoupling the supply in various ways and it just >>>>>>>> doesn't seem to be having any effect on the oscillation. I thought that >>>>>>>> perhaps it might be due to the higher gain of the LM13700 at the higher >>>>>>>> supply voltage, but today I tried disconnecting the signal going into >>>>>>>> the control current input, which caused the frequency of the >>>>>>>> oscillation >>>>>>>> to go up to about 80 Khz. With the control current input disconnected, >>>>>>>> the gain should theoretically be zero... >>>>>>>> >>>>>>> It's really unlikely to be a bypassing issue at 150 Hz. >>>>>>> >>>>>>> Motorboating is due to accumulated phase _lead_ at low frequency, >>>>>>> rather than lag at high frequency. Only happens in AC-coupled >>>>>>> stages, so if your amp is DC-coupled, that isn't it. >>>>>> Here's a link to the relevant portion of the schematic, though it >>>>>> isn't completed yet: >>>>>> http://i227.photobucket.com/albums/dd240/bitrex2007/DualVCA.jpg >>>>>> >>>>>> Signal comes in from the upper left and is divided down by the 100k >>>>>> and 100 ohm resistors, and the control signal comes in from the bottom >>>>>> right and is made into two signals 180 degrees out of phase for the >>>>>> two VCAs. >>>>>> Output will be at the upper right. Right now for testing I just have >>>>>> one current source/VCA/output pair hooked up on the breadboard using >>>>>> signal generators as inputs. The only AC coupling will be at the >>>>>> input and output of the circuit. >>>>>> >>>>>> >>>>>>> It may also be that your feedback loop becomes unstable when the very >>>>>>> low operating current makes its gain bandwidth get really really low. >>>>>>> Why not post a schematic? >>>>>> Since the LM13700 is being used as a VCA there's no feedback loop >>>>>> around it, exactly; but I think I understand your point - if the gain >>>>>> is very low the bandwidth will be wide and make the amp prone to >>>>>> oscillation. There's bandwidth limiting in the input stage of the >>>>>> actual circuit but for test purposes I was just driving it off a >>>>>> signal generator. I'll try putting a low pass filter on the input of >>>>>> the test circuit and see if that helps. >>>>>> >>>>> Besides feedback, aren't you supposed to tie the diode tap to the >>>>> positive supply via a resistor? >>>>> >>>> I haven't decided what to do with those, yet. The trade off for the >>>> improvement in distortion is that those diodes are supposed to be run at >>>> a relatively high current to get it...in this application (run from a >>>> battery) I don't know if the improvement in distortion is worth the >>>> extra couple of mA. Then again it's only about as much as an extra op >>>> amp section to linearize each LM13700 and I'm already using 10 sections >>>> in the design...eh. >>>> >>>> >>>> >>>> >>>> >>> Never mind that your loop gain is all over the place ?:-) >>> >>> ...Jim Thompson >> Even with the smiley, such a comment from an experienced designer makes >> me wonder if it's just a joke about VCAs being inherently distorting or >> if I've done something blindingly wrong! :( > > I don't know that VCA's are "inherently distorting", but the loop gain > variation can be wild, particularly at very low currents. > > Download the LoopGain checker from my website and simulate. > > ...Jim Thompson Thanks! I will definitely do that. I'm curious as to the origin of these loop gain variations - in an LM13700 the gain is controlled by sinking a current into a base current compensated current mirror, the other half of which is connected to the input differential pair to control the gain. It would _seem_ that there's little room for things to go wrong...
From: Jim Thompson on 26 Apr 2010 17:59 On Mon, 26 Apr 2010 17:49:28 -0400, Bitrex <bitrex(a)de.lete.earthlink.net> wrote: >Jim Thompson wrote: >> On Mon, 26 Apr 2010 06:07:04 -0400, Bitrex >> <bitrex(a)de.lete.earthlink.net> wrote: >> >>> Jim Thompson wrote: >>>> On Sun, 25 Apr 2010 19:57:59 -0400, Bitrex >>>> <bitrex(a)de.lete.earthlink.net> wrote: >>>> >>>>> Joerg wrote: >>>>>> Bitrex wrote: >>>>>>> Phil Hobbs wrote: >>>>>>>> On 4/24/2010 6:25 PM, Bitrex wrote: >>>>>>>>> Phil Hobbs wrote: >>>>>>>>>> On 4/24/2010 2:23 AM, Bitrex wrote: >>>>>>>>>>> I'm using an LM13700 as a VCA in a 9V battery powered circuit, >>>>>>>>>>> using an >>>>>>>>>>> opamp buffered virtual ground. The circuit works fine when powered >>>>>>>>>>> off >>>>>>>>>>> the battery, but I wanted to test it using different supply voltages. >>>>>>>>>>> When I hooked up the supply to the 15 volt breaboard power supply, I >>>>>>>>>>> noticed that the LM13700 was oscillating at about 150 Hz when >>>>>>>>>>> there was >>>>>>>>>>> no input signal - the oscillation stops when signal is applied and >>>>>>>>>>> the >>>>>>>>>>> VCA turns on. I couldn't seem to get it to settle down with the usual >>>>>>>>>>> supply bypass capacitors, whatever values I tried. Maybe it's just a >>>>>>>>>>> consequence of sloppy breadboard layout? The other ICs in the circuit >>>>>>>>>>> (mostly TL084s) work OK regardless of supply choice. Suggestions >>>>>>>>>>> would >>>>>>>>>>> be appreciated! >>>>>>>>>> Motorboating? >>>>>>>>>> >>>>>>>>>> Cheers >>>>>>>>>> >>>>>>>>>> Phil Hobbs >>>>>>>>>> >>>>>>>>> Maybe, but I've tried decoupling the supply in various ways and it just >>>>>>>>> doesn't seem to be having any effect on the oscillation. I thought that >>>>>>>>> perhaps it might be due to the higher gain of the LM13700 at the higher >>>>>>>>> supply voltage, but today I tried disconnecting the signal going into >>>>>>>>> the control current input, which caused the frequency of the >>>>>>>>> oscillation >>>>>>>>> to go up to about 80 Khz. With the control current input disconnected, >>>>>>>>> the gain should theoretically be zero... >>>>>>>>> >>>>>>>> It's really unlikely to be a bypassing issue at 150 Hz. >>>>>>>> >>>>>>>> Motorboating is due to accumulated phase _lead_ at low frequency, >>>>>>>> rather than lag at high frequency. Only happens in AC-coupled >>>>>>>> stages, so if your amp is DC-coupled, that isn't it. >>>>>>> Here's a link to the relevant portion of the schematic, though it >>>>>>> isn't completed yet: >>>>>>> http://i227.photobucket.com/albums/dd240/bitrex2007/DualVCA.jpg >>>>>>> >>>>>>> Signal comes in from the upper left and is divided down by the 100k >>>>>>> and 100 ohm resistors, and the control signal comes in from the bottom >>>>>>> right and is made into two signals 180 degrees out of phase for the >>>>>>> two VCAs. >>>>>>> Output will be at the upper right. Right now for testing I just have >>>>>>> one current source/VCA/output pair hooked up on the breadboard using >>>>>>> signal generators as inputs. The only AC coupling will be at the >>>>>>> input and output of the circuit. >>>>>>> >>>>>>> >>>>>>>> It may also be that your feedback loop becomes unstable when the very >>>>>>>> low operating current makes its gain bandwidth get really really low. >>>>>>>> Why not post a schematic? >>>>>>> Since the LM13700 is being used as a VCA there's no feedback loop >>>>>>> around it, exactly; but I think I understand your point - if the gain >>>>>>> is very low the bandwidth will be wide and make the amp prone to >>>>>>> oscillation. There's bandwidth limiting in the input stage of the >>>>>>> actual circuit but for test purposes I was just driving it off a >>>>>>> signal generator. I'll try putting a low pass filter on the input of >>>>>>> the test circuit and see if that helps. >>>>>>> >>>>>> Besides feedback, aren't you supposed to tie the diode tap to the >>>>>> positive supply via a resistor? >>>>>> >>>>> I haven't decided what to do with those, yet. The trade off for the >>>>> improvement in distortion is that those diodes are supposed to be run at >>>>> a relatively high current to get it...in this application (run from a >>>>> battery) I don't know if the improvement in distortion is worth the >>>>> extra couple of mA. Then again it's only about as much as an extra op >>>>> amp section to linearize each LM13700 and I'm already using 10 sections >>>>> in the design...eh. >>>>> >>>>> >>>>> >>>>> >>>>> >>>> Never mind that your loop gain is all over the place ?:-) >>>> >>>> ...Jim Thompson >>> Even with the smiley, such a comment from an experienced designer makes >>> me wonder if it's just a joke about VCAs being inherently distorting or >>> if I've done something blindingly wrong! :( >> >> I don't know that VCA's are "inherently distorting", but the loop gain >> variation can be wild, particularly at very low currents. >> >> Download the LoopGain checker from my website and simulate. >> >> ...Jim Thompson > >Thanks! I will definitely do that. I'm curious as to the origin of >these loop gain variations - in an LM13700 the gain is controlled by >sinking a current into a base current compensated current mirror, the >other half of which is connected to the input differential pair to >control the gain. It would _seem_ that there's little room for things >to go wrong... The gain varies ?:-) ...Jim Thompson -- | James E.Thompson, CTO | mens | | Analog Innovations, Inc. | et | | Analog/Mixed-Signal ASIC's and Discrete Systems | manus | | Phoenix, Arizona 85048 Skype: Contacts Only | | | Voice:(480)460-2350 Fax: Available upon request | Brass Rat | | E-mail Icon at http://www.analog-innovations.com | 1962 | The only thing bipartisan in this country is hypocrisy
From: Bitrex on 26 Apr 2010 18:33
Jim Thompson wrote: > On Mon, 26 Apr 2010 17:49:28 -0400, Bitrex > <bitrex(a)de.lete.earthlink.net> wrote: > >> Jim Thompson wrote: >>> On Mon, 26 Apr 2010 06:07:04 -0400, Bitrex >>> <bitrex(a)de.lete.earthlink.net> wrote: >>> >>>> Jim Thompson wrote: >>>>> On Sun, 25 Apr 2010 19:57:59 -0400, Bitrex >>>>> <bitrex(a)de.lete.earthlink.net> wrote: >>>>> >>>>>> Joerg wrote: >>>>>>> Bitrex wrote: >>>>>>>> Phil Hobbs wrote: >>>>>>>>> On 4/24/2010 6:25 PM, Bitrex wrote: >>>>>>>>>> Phil Hobbs wrote: >>>>>>>>>>> On 4/24/2010 2:23 AM, Bitrex wrote: >>>>>>>>>>>> I'm using an LM13700 as a VCA in a 9V battery powered circuit, >>>>>>>>>>>> using an >>>>>>>>>>>> opamp buffered virtual ground. The circuit works fine when powered >>>>>>>>>>>> off >>>>>>>>>>>> the battery, but I wanted to test it using different supply voltages. >>>>>>>>>>>> When I hooked up the supply to the 15 volt breaboard power supply, I >>>>>>>>>>>> noticed that the LM13700 was oscillating at about 150 Hz when >>>>>>>>>>>> there was >>>>>>>>>>>> no input signal - the oscillation stops when signal is applied and >>>>>>>>>>>> the >>>>>>>>>>>> VCA turns on. I couldn't seem to get it to settle down with the usual >>>>>>>>>>>> supply bypass capacitors, whatever values I tried. Maybe it's just a >>>>>>>>>>>> consequence of sloppy breadboard layout? The other ICs in the circuit >>>>>>>>>>>> (mostly TL084s) work OK regardless of supply choice. Suggestions >>>>>>>>>>>> would >>>>>>>>>>>> be appreciated! >>>>>>>>>>> Motorboating? >>>>>>>>>>> >>>>>>>>>>> Cheers >>>>>>>>>>> >>>>>>>>>>> Phil Hobbs >>>>>>>>>>> >>>>>>>>>> Maybe, but I've tried decoupling the supply in various ways and it just >>>>>>>>>> doesn't seem to be having any effect on the oscillation. I thought that >>>>>>>>>> perhaps it might be due to the higher gain of the LM13700 at the higher >>>>>>>>>> supply voltage, but today I tried disconnecting the signal going into >>>>>>>>>> the control current input, which caused the frequency of the >>>>>>>>>> oscillation >>>>>>>>>> to go up to about 80 Khz. With the control current input disconnected, >>>>>>>>>> the gain should theoretically be zero... >>>>>>>>>> >>>>>>>>> It's really unlikely to be a bypassing issue at 150 Hz. >>>>>>>>> >>>>>>>>> Motorboating is due to accumulated phase _lead_ at low frequency, >>>>>>>>> rather than lag at high frequency. Only happens in AC-coupled >>>>>>>>> stages, so if your amp is DC-coupled, that isn't it. >>>>>>>> Here's a link to the relevant portion of the schematic, though it >>>>>>>> isn't completed yet: >>>>>>>> http://i227.photobucket.com/albums/dd240/bitrex2007/DualVCA.jpg >>>>>>>> >>>>>>>> Signal comes in from the upper left and is divided down by the 100k >>>>>>>> and 100 ohm resistors, and the control signal comes in from the bottom >>>>>>>> right and is made into two signals 180 degrees out of phase for the >>>>>>>> two VCAs. >>>>>>>> Output will be at the upper right. Right now for testing I just have >>>>>>>> one current source/VCA/output pair hooked up on the breadboard using >>>>>>>> signal generators as inputs. The only AC coupling will be at the >>>>>>>> input and output of the circuit. >>>>>>>> >>>>>>>> >>>>>>>>> It may also be that your feedback loop becomes unstable when the very >>>>>>>>> low operating current makes its gain bandwidth get really really low. >>>>>>>>> Why not post a schematic? >>>>>>>> Since the LM13700 is being used as a VCA there's no feedback loop >>>>>>>> around it, exactly; but I think I understand your point - if the gain >>>>>>>> is very low the bandwidth will be wide and make the amp prone to >>>>>>>> oscillation. There's bandwidth limiting in the input stage of the >>>>>>>> actual circuit but for test purposes I was just driving it off a >>>>>>>> signal generator. I'll try putting a low pass filter on the input of >>>>>>>> the test circuit and see if that helps. >>>>>>>> >>>>>>> Besides feedback, aren't you supposed to tie the diode tap to the >>>>>>> positive supply via a resistor? >>>>>>> >>>>>> I haven't decided what to do with those, yet. The trade off for the >>>>>> improvement in distortion is that those diodes are supposed to be run at >>>>>> a relatively high current to get it...in this application (run from a >>>>>> battery) I don't know if the improvement in distortion is worth the >>>>>> extra couple of mA. Then again it's only about as much as an extra op >>>>>> amp section to linearize each LM13700 and I'm already using 10 sections >>>>>> in the design...eh. >>>>>> >>>>>> >>>>>> >>>>>> >>>>>> >>>>> Never mind that your loop gain is all over the place ?:-) >>>>> >>>>> ...Jim Thompson >>>> Even with the smiley, such a comment from an experienced designer makes >>>> me wonder if it's just a joke about VCAs being inherently distorting or >>>> if I've done something blindingly wrong! :( >>> I don't know that VCA's are "inherently distorting", but the loop gain >>> variation can be wild, particularly at very low currents. >>> >>> Download the LoopGain checker from my website and simulate. >>> >>> ...Jim Thompson >> Thanks! I will definitely do that. I'm curious as to the origin of >> these loop gain variations - in an LM13700 the gain is controlled by >> sinking a current into a base current compensated current mirror, the >> other half of which is connected to the input differential pair to >> control the gain. It would _seem_ that there's little room for things >> to go wrong... > > The gain varies ?:-) > > ...Jim Thompson Well, right, I guess I thought that you were implying that it varies _unpredictably_ at low currents. I think I see how that could happen, in a perfect world the transconductance would just be I_0/2*VT, where I_0 is the reference current...but of course even the Wilson current mirror in the IC is less than perfect at low currents. |