LT1054 Voltage Inverter fails to start up with no load

Question

Hello, 
 I'm using LT1054 to produce -12V from +12V using the "Basic Voltage Inverter" circuit from "Typical Applications" on the data sheet. 
 I have disconnected the load from the -12V output from the LT1054. 
 
 The issue I have is that the LT1054 fails to start up properly when 12V is applied. 
 The OSC pin produces an irregular waveform, not the expected sawtooth. The VREF pun does not produce a stable output. VOUT is not stable and is about -4V on average. 
 However, if I momentarily touch FB/SHDN with a DMM probe, the device switches to normal operation. The output is -12V, OSC produces a sawtooth and VREF is stable. 
 
 There is a warning in the data sheet about allowing VOUT to drift positive due to a positive supply referenced load. However, at no time does VOUT go positive. 
 Can anybody help me understand this issue, please?

legeyt · Accepted Answer

Hello, 
 I faced the same issue - LT1054 was not starting up in some cases. I will detail below what I observed and what worked for me. 
 I am using the LT1054 to convert +12V to -12V. The output is driving the input of a TPS7A30 negative voltage regulator. The ultimate load is a pair of ADA4637 op-amps (14mA total supply current). There is also a positive regulator in this circuit but I enable it with a slow R-C to ensure that it starts after the negative regulator so that there is no chance of the output of the LT1054 being pulled positive (I verified this with a scope). 
 I first attempted to run this circuit with a flying capacitor of 4.4&micro;F and an output capacitor of 44&micro;F. These are 0805 ceramic capacitors (2 x 2.2&micro;F and 2 x 22&micro;F). There is 2x4.7&micro;F on the input (shared with the positive regulator, also ceramic). Under this circumstance, the LT1054 would start normally when the power supply ramped up slowly (~10ms) but when the power supply ramped quickly (~10&micro;s) it would fall into a bad state as described by the previous poster. After reading the above post I found that mine would also sometimes regulate normally after touching the FB pin with a multimeter probe (when measuring relative to ground). I assume this is related to the 'reset' described in the datasheet. In my case the device was drawing busts of current at around 1kHz but never enough to get past about -2V at the output. It was easy to tell when the device failed to start up because I could audibly hear the ceramic capacitors singing. 
 I tried adding the feedback network (tuned for -12V) but this did not have any effect. 
 In the end I increased the capacitors to 9.1&micro;F input and 88&micro;F output to more closely match the datasheet values of 10&micro;F and 100&micro;F and now the device starts up with both slow and fast supply ramping. I find the datasheet misleading in this regard as it says on page 14: 
 "While the exact values of CIN and COUT are noncritical, good-quality low-ESR capacitors, such as solid tantalum, are necessary to minimize voltage losses at high currents." 
 It also says lower on the page: 
 "For best results, this ratio should be approximately 1:10." 
 There is no mention that I found of a minimum value for the output capacitor - if the device truly requires 100&micro;F output to function properly it would seem to me that this should be stated clearly rather than saying "the exact values ... are noncritical")

legeyt · Answer

Hello, 
 I faced the same issue - LT1054 was not starting up in some cases. I will detail below what I observed and what worked for me. 
 I am using the LT1054 to convert +12V to -12V. The output is driving the input of a TPS7A30 negative voltage regulator. The ultimate load is a pair of ADA4637 op-amps (14mA total supply current). There is also a positive regulator in this circuit but I enable it with a slow R-C to ensure that it starts after the negative regulator so that there is no chance of the output of the LT1054 being pulled positive (I verified this with a scope). 
 I first attempted to run this circuit with a flying capacitor of 4.4&micro;F and an output capacitor of 44&micro;F. These are 0805 ceramic capacitors (2 x 2.2&micro;F and 2 x 22&micro;F). There is 2x4.7&micro;F on the input (shared with the positive regulator, also ceramic). Under this circumstance, the LT1054 would start normally when the power supply ramped up slowly (~10ms) but when the power supply ramped quickly (~10&micro;s) it would fall into a bad state as described by the previous poster. After reading the above post I found that mine would also sometimes regulate normally after touching the FB pin with a multimeter probe (when measuring relative to ground). I assume this is related to the 'reset' described in the datasheet. In my case the device was drawing busts of current at around 1kHz but never enough to get past about -2V at the output. It was easy to tell when the device failed to start up because I could audibly hear the ceramic capacitors singing. 
 I tried adding the feedback network (tuned for -12V) but this did not have any effect. 
 In the end I increased the capacitors to 9.1&micro;F input and 88&micro;F output to more closely match the datasheet values of 10&micro;F and 100&micro;F and now the device starts up with both slow and fast supply ramping. I find the datasheet misleading in this regard as it says on page 14: 
 "While the exact values of CIN and COUT are noncritical, good-quality low-ESR capacitors, such as solid tantalum, are necessary to minimize voltage losses at high currents." 
 It also says lower on the page: 
 "For best results, this ratio should be approximately 1:10." 
 There is no mention that I found of a minimum value for the output capacitor - if the device truly requires 100&micro;F output to function properly it would seem to me that this should be stated clearly rather than saying "the exact values ... are noncritical"

Electronic Design Services - PartnerZone

Products Mentioned

LT1054 Voltage Inverter fails to start up with no load